{"as_of":"2026-08-11T03:30:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:d506101bd6934035e925a2e6e981f72a6ae6a32e204a11c2b02b12f8848381a7","coverage":[{"denominator":0,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":100,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":100,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-10T06:31:04.303077+00:00","state":"measured"},{"denominator":125,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":100,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-08T21:06:55.819737Z","state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":1,"source":"pith","source_observed_at":"2026-08-05T02:28:24.338817Z","state":"measured"}],"external_citation_measurements":[{"count":12,"observed_at":"2026-08-05T02:28:24.338817Z","source":"pith"}],"inbound":[{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-08T21:06:55.819737Z","title":"Reasoning models don’t always say what they think","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2502.05242","last_updated":"2026-05-27T09:11:42Z","snapshot_observed_at":"2026-08-08T20:58:31.381100Z","submitted_at":"2025-02-07T13:25:33Z","title":"Beyond External Monitors: Enhancing Transparency of Large Language Models for Easier Monitoring","version":3},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-08T21:06:55.819737Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2502.05242"},"observation_digest":"sha256:dd7fef6fc21c31b32e10257bfb4fed7449a17a55f8e050295000fb2f49178743","observation_id":"b0d84af8-53d5-4908-8fbe-29a7d337e092","resolution":{"observed_at":"2026-08-08T21:06:55.819737Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T14:52:40.853592Z","title":"Reasoning models don’t always say what they think.arXiv preprint arXiv:2505.05410, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2505.17407","last_updated":"2025-05-23T02:46:18Z","snapshot_observed_at":"2026-08-08T01:26:05.514174Z","submitted_at":"2025-05-23T02:46:18Z","title":"Language Matters: How Do Multilingual Input and Reasoning Paths Affect Large Reasoning Models?","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-07T14:52:40.853592Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2505.17407"},"observation_digest":"sha256:9c24c229c2cc16178f35ccd1564fb283cef9389622ec115b7a8cb1d2e5e86626","observation_id":"e584c35e-928a-4b1c-b8b7-db77b142f3fb","resolution":{"observed_at":"2026-08-07T14:52:40.853592Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T14:31:48.472560Z","title":"Reasoning models don't always say what they think","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2505.21537","last_updated":"2025-05-24T09:07:13Z","snapshot_observed_at":"2026-08-08T02:37:32.325481Z","submitted_at":"2025-05-24T09:07:13Z","title":"OpenReview Should be Protected and Leveraged as a Community Asset for Research in the Era of Large Language Models","version":1},"reference_index":158,"source":"arxiv_source","source_observed_at":"2026-08-07T14:31:48.472560Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2505.21537"},"observation_digest":"sha256:2a8d58d70a8e756a38c82b473dee97692ce9d671d18a0e83b9e85fae8ceab775","observation_id":"eab382ac-af4b-45d1-a118-f797cd15a0b5","resolution":{"observed_at":"2026-08-07T14:31:48.472560Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T12:59:21.225815Z","title":"https://doi.org/10.48550/arXiv.2505.05410","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2505.22987","last_updated":"2025-07-17T20:04:13Z","snapshot_observed_at":"2026-08-08T02:09:32.171416Z","submitted_at":"2025-05-29T01:51:20Z","title":"Strategic Reflectivism In Intelligent Systems","version":2},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-07T12:59:21.225815Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2505.22987"},"observation_digest":"sha256:5253459b9768584029bfe3ebcf0c963ecc7478611e3a6e79face330e090c9d00","observation_id":"e7f3d973-79fb-47d3-8a48-616c1f3674ff","resolution":{"observed_at":"2026-08-07T12:59:21.225815Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T12:02:40.820858Z","title":"Reasoning models don't always say what they think","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2506.00845","last_updated":"2025-08-17T12:17:48Z","snapshot_observed_at":"2026-08-07T16:09:18.251031Z","submitted_at":"2025-06-01T05:39:56Z","title":"Generalizable LLM Learning of Graph Synthetic Data with Post-training Alignment","version":3},"reference_index":9,"source":"arxiv_source","source_observed_at":"2026-08-07T12:02:40.820858Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.00845"},"observation_digest":"sha256:47a329fc4a500773075d4b2a74922c0057dc941f689743a7540af4e53b435b45","observation_id":"8fea7934-502b-4000-97ef-23decd091a22","resolution":{"observed_at":"2026-08-07T12:02:40.820858Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T11:21:49.677629Z","title":"Reasoning models don’t always say what they think,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2506.02739","last_updated":"2025-06-03T10:53:29Z","snapshot_observed_at":"2026-08-09T13:20:41.424260Z","submitted_at":"2025-06-03T10:53:29Z","title":"Why do AI agents communicate in human language?","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-07T11:21:49.677629Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.02739"},"observation_digest":"sha256:aaa522ea4efca5c55501a8cb712cd4b2efbe58bfeda17294fe5f261d70a46b84","observation_id":"d658fca0-f320-4b32-9929-f9116c6da8c0","resolution":{"observed_at":"2026-08-07T11:21:49.677629Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2506.06941","last_updated":"2025-11-20T00:19:24Z","snapshot_observed_at":"2026-08-09T21:19:24.140229Z","submitted_at":"2025-06-07T22:42:29Z","title":"The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity","version":3},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-05-15T16:10:31.440921Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.06941"},"observation_digest":"sha256:6a71b38928361c03c797aafcd1fa2753c5225fc7e9f76bafc3909979bd452e9e","observation_id":"3897f119-b35d-4837-b2a0-23f98bbac994","resolution":{"observed_at":"2026-05-15T16:10:31.664485Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T05:14:45.045847Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2506.08446","last_updated":"2025-06-10T04:44:28Z","snapshot_observed_at":"2026-08-07T09:15:20.058485Z","submitted_at":"2025-06-10T04:44:28Z","title":"A Survey on Large Language Models for Mathematical Reasoning","version":1},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-07T05:14:45.045847Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.08446"},"observation_digest":"sha256:69e8fda9cbc590bd4e608339c22048d003fcdef0ceb9d1849390dd325404d61e","observation_id":"cf7e1040-bede-4c42-b203-413bc87f8b2f","resolution":{"observed_at":"2026-08-07T05:14:45.045847Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-07T05:01:08.327546Z","title":"Bowman, Jan Leike, Jared Kaplan, and Ethan Perez","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2506.09038","last_updated":"2025-06-10T17:57:30Z","snapshot_observed_at":"2026-08-10T22:41:21.434916Z","submitted_at":"2025-06-10T17:57:30Z","title":"AbstentionBench: Reasoning LLMs Fail on Unanswerable Questions","version":1},"reference_index":83,"source":"pdf_text","source_observed_at":"2026-08-07T05:01:08.327546Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.09038"},"observation_digest":"sha256:cc50c938dd7d90aeeace796c73848b5db54984d0d344ed2d66c42759b5de78d1","observation_id":"0df62072-9738-438c-969c-a79048aecda3","resolution":{"observed_at":"2026-08-07T05:01:08.327546Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T22:03:51.958119Z","title":"Reasoning models don’t always say what they think.arXiv preprint arXiv:2505.05410, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2506.22777","last_updated":"2025-07-13T15:36:35Z","snapshot_observed_at":"2026-08-10T05:25:44.170625Z","submitted_at":"2025-06-28T06:37:10Z","title":"Teaching Models to Verbalize Reward Hacking in Chain-of-Thought Reasoning","version":2},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-06T22:03:51.958119Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.22777"},"observation_digest":"sha256:e671d662cc0c220f7fd5c136aa2eaa930f996451d80da2615322968ae7e06e25","observation_id":"2170fee5-fc74-4209-a415-4a639f0c0b89","resolution":{"observed_at":"2026-08-06T22:03:51.958119Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2506.22832","last_updated":"2026-04-09T11:44:23Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2025-06-28T09:53:17Z","title":"Listener-Rewarded Thinking in VLMs for Image Preferences","version":3},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-05-19T07:38:52.273903Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.22832"},"observation_digest":"sha256:435dca31f5d93c0fb604dbab0c4428f3a0040bccf3c9ce747619c48ac0577918","observation_id":"536ecd38-a839-4c65-af8c-9b27a028b125","resolution":{"observed_at":"2026-05-19T07:42:09.240618Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T21:37:52.644431Z","title":"Bowman, Jan Leike, Jared Kaplan, and Ethan Perez","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2506.23678","last_updated":"2025-06-30T10:00:43Z","snapshot_observed_at":"2026-08-10T20:57:42.982269Z","submitted_at":"2025-06-30T10:00:43Z","title":"Interactive Reasoning: Visualizing and Controlling Chain-of-Thought Reasoning in Large Language Models","version":1},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-06T21:37:52.644431Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2506.23678"},"observation_digest":"sha256:9a6c969be0b8de36179c2055435c34cf11fc7ec961304067f033977ec8ea055e","observation_id":"465cd223-356b-406d-9dee-178694ad797e","resolution":{"observed_at":"2026-08-06T21:37:52.644431Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T20:25:53.363109Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2507.02799","last_updated":"2025-07-03T17:01:53Z","snapshot_observed_at":"2026-08-06T20:18:22.609860Z","submitted_at":"2025-07-03T17:01:53Z","title":"Is Reasoning All You Need? Probing Bias in the Age of Reasoning Language Models","version":1},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-06T20:25:53.363109Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2507.02799"},"observation_digest":"sha256:19ae1b229aa9854741eccc79c3de4add091cfc5756ff84c3fbe3e4cec9cc7512","observation_id":"40f0cd15-8704-4053-a0ff-37067452c6a2","resolution":{"observed_at":"2026-08-06T20:25:53.363109Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T20:18:15.499693Z","title":"Chen et al., ‘Reasoning Models Don’t Always Say What They Think’, May 08, 2025, arXiv: arXiv:2505.05410","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2507.03409","last_updated":"2025-07-04T09:16:11Z","snapshot_observed_at":"2026-08-08T10:46:44.107781Z","submitted_at":"2025-07-04T09:16:11Z","title":"Lessons from a Chimp: AI \"Scheming\" and the Quest for Ape Language","version":1},"reference_index":53,"source":"pdf_text","source_observed_at":"2026-08-06T20:18:15.499693Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2507.03409"},"observation_digest":"sha256:c683de8366197470ad5dc3ac37bd94c15e68278529eaeb490992e5ba695f17a1","observation_id":"88018fc3-e8f8-4346-9d24-ad3dee5f9bf8","resolution":{"observed_at":"2026-08-06T20:18:15.499693Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T16:12:47.648456Z","title":"Reasoning models don’t always say what they think","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2507.14293","last_updated":"2025-07-18T18:06:27Z","snapshot_observed_at":"2026-08-06T15:57:20.673152Z","submitted_at":"2025-07-18T18:06:27Z","title":"WebGuard: Building a Generalizable Guardrail for Web Agents","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-06T16:12:47.648456Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2507.14293"},"observation_digest":"sha256:ae3e504992b01d6d32e44e9aebe54e2b1ddc9b014319dce1ce5aa439c61abfb9","observation_id":"caa0589b-8752-4308-b471-8fef84c0721a","resolution":{"observed_at":"2026-08-06T16:12:47.648456Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T15:20:10.865469Z","title":"Reasoning models don't always say what they think","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2507.16331","last_updated":"2026-07-03T09:17:03Z","snapshot_observed_at":"2026-08-07T04:15:41.914807Z","submitted_at":"2025-07-22T08:13:01Z","title":"Re:Form -- Reducing Human Annotations in Scalable Formal Software Verification with RL in LLMs: A Preliminary Study on Dafny","version":4},"reference_index":9,"source":"arxiv_source","source_observed_at":"2026-08-06T15:20:10.865469Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2507.16331"},"observation_digest":"sha256:e3d50651f8930486958280764d1e15d1fcc451c272adfbada8fa48ac4059022f","observation_id":"5795bd41-e36f-4944-a6bc-19e2b558e834","resolution":{"observed_at":"2026-08-06T15:20:10.865469Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-06T11:06:07.060768Z","title":"Reasoning Models Don’t Always Say What They Think [Internet]","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2507.23146","last_updated":"2026-07-22T18:38:53Z","snapshot_observed_at":"2026-08-06T11:06:05.284236Z","submitted_at":"2025-07-30T22:48:34Z","title":"Lightweight Language Models are Prone to Reasoning Errors for Complex Computational Phenotyping Tasks","version":4},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-06T11:06:07.060768Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2507.23146"},"observation_digest":"sha256:79617bacaf4955c797f5a43b44410f204d8fd4ec56ac058ceb83b91b76f23242","observation_id":"ef73b372-82cc-494e-bfc6-1fdb4ff1aad1","resolution":{"observed_at":"2026-08-06T11:06:07.060768Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T23:01:59.326685Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2508.06017","last_updated":"2025-08-08T05:04:47Z","snapshot_observed_at":"2026-08-08T10:44:52.352035Z","submitted_at":"2025-08-08T05:04:47Z","title":"Position: Intelligent Coding Systems Should Write Programs with Justifications","version":1},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-05T23:01:59.326685Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2508.06017"},"observation_digest":"sha256:5ae764c6ff047e1eabd5e7411ba006a8666cb7f6d3a91a31d11ced1291c2562d","observation_id":"640802df-da7a-4574-9bb4-b6d7ea69a610","resolution":{"observed_at":"2026-08-05T23:01:59.326685Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T15:53:48.111145Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2508.19461","last_updated":"2025-08-26T22:29:31Z","snapshot_observed_at":"2026-08-07T19:17:20.900756Z","submitted_at":"2025-08-26T22:29:31Z","title":"Reliable Weak-to-Strong Monitoring of LLM Agents","version":1},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-05T15:53:48.111145Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2508.19461"},"observation_digest":"sha256:d51e9a65ce3a31d1fec3159c2699d1e2473feac769b3f0c94923f220685d8460","observation_id":"acfa9999-7359-4316-860a-4b79b5086acf","resolution":{"observed_at":"2026-08-05T15:53:48.111145Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-04T19:12:05.814286Z","title":"Bowman, Jan Leike, Jared Kaplan, and Ethan Perez","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2509.09396","last_updated":"2025-09-11T12:25:41Z","snapshot_observed_at":"2026-08-09T11:34:45.899331Z","submitted_at":"2025-09-11T12:25:41Z","title":"LLMs Don't Know Their Own Decision Boundaries: The Unreliability of Self-Generated Counterfactual Explanations","version":1},"reference_index":7,"source":"arxiv_source","source_observed_at":"2026-08-04T19:12:05.814286Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2509.09396"},"observation_digest":"sha256:61f73fac039e3309faf52bc5a72c985ccd576e1752d60ab5493e235eb41ce9e7","observation_id":"e706203d-4698-45ef-9caf-4a22f187b0d5","resolution":{"observed_at":"2026-08-04T19:12:05.814286Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-04T17:40:08.174781Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2509.10723","last_updated":"2025-09-12T22:26:31Z","snapshot_observed_at":"2026-08-04T17:40:07.040846Z","submitted_at":"2025-09-12T22:26:31Z","title":"Dark Patterns Meet GUI Agents: LLM Agent Susceptibility to Manipulative Interfaces and the Role of Human Oversight","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-04T17:40:08.174781Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2509.10723"},"observation_digest":"sha256:58800dbd45c76c484ae398a2841b89524beaacf4fcc026f2fbcc9cd3fffe40bd","observation_id":"3653af64-4892-4a1b-a832-592267de95d2","resolution":{"observed_at":"2026-08-04T17:40:08.174781Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2509.21361","last_updated":"2026-04-23T15:39:44Z","snapshot_observed_at":"2026-07-06T22:30:46.129271Z","submitted_at":"2025-09-21T14:38:17Z","title":"Context Is What You Need: The Maximum Effective Context Window for Real World Limits of LLMs","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-05-18T14:51:42.879926Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2509.21361"},"observation_digest":"sha256:2f2f3c79f5a8846e82da1857286dbda16e6189149ce7cfe0ef2ebd1a49063046","observation_id":"33013157-cb8a-442f-9429-a3e535258762","resolution":{"observed_at":"2026-05-18T14:52:41.098958Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-04T10:40:42.937652Z","title":"Bowman, Jan Leike, Jared Kaplan, and Ethan Perez","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2510.09278","last_updated":"2026-06-28T16:27:56Z","snapshot_observed_at":"2026-08-08T22:59:31.970086Z","submitted_at":"2025-10-10T11:21:09Z","title":"CLARity: Reasoning Consistency Alone Can Teach Reinforced Experts","version":2},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-08-04T10:40:42.937652Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2510.09278"},"observation_digest":"sha256:ead42df1e6edb27fff92414691a357b10db6ba80b247030383e657b019d6e1ac","observation_id":"399715de-4411-4b6e-a0b8-7346bfde06b6","resolution":{"observed_at":"2026-08-04T10:40:42.937652Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-04T09:16:39.944551Z","title":"11 Preprint","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2510.16380","last_updated":"2026-06-10T21:26:48Z","snapshot_observed_at":"2026-08-08T03:07:18.068977Z","submitted_at":"2025-10-18T07:34:31Z","title":"MoReBench: Evaluating Procedural and Pluralistic Moral Reasoning in Language Models, More than Outcomes","version":2},"reference_index":2025,"source":"pdf_text","source_observed_at":"2026-08-04T09:16:39.944551Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2510.16380"},"observation_digest":"sha256:45ee33ae45ce50d6562b68fe55dac75cb4ab37c40b0a16603cb432df2de53d80","observation_id":"16b83c19-f98d-4cdb-8de1-2f7b2e26fe91","resolution":{"observed_at":"2026-08-04T09:16:39.944551Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2510.18814","last_updated":"2026-07-05T03:44:20Z","snapshot_observed_at":"2026-08-04T08:53:03.599438Z","submitted_at":"2025-10-21T17:15:56Z","title":"A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-05-18T05:11:29.205366Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2510.18814"},"observation_digest":"sha256:4a97af34ed659c9df42cbfcfc43630de182597183fd3669543617c940dcd73ec","observation_id":"b0d2b242-9c7e-42cc-8bd1-5d10b8716736","resolution":{"observed_at":"2026-05-18T05:12:23.661588Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2510.18814","last_updated":"2026-07-05T03:44:20Z","snapshot_observed_at":"2026-08-04T08:53:03.599438Z","submitted_at":"2025-10-21T17:15:56Z","title":"A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning","version":3},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-05-21T20:06:16.172916Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2510.18814"},"observation_digest":"sha256:790f2b283674adea6f1692db3c7d379d68fd4fce2abc63c8e19e837502f8eeca","observation_id":"b41a6e13-c940-409a-8066-8b6103ea7860","resolution":{"observed_at":"2026-05-21T20:10:34.943343Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-04T08:53:05.547904Z","title":"Reasoning models don’t always say what they think.arXiv preprint arXiv:2505.05410,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2510.18814","last_updated":"2026-07-05T03:44:20Z","snapshot_observed_at":"2026-08-04T08:53:03.599438Z","submitted_at":"2025-10-21T17:15:56Z","title":"A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning","version":4},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-04T08:53:05.547904Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2510.18814"},"observation_digest":"sha256:7e7266b20d13c434d2169ee9cf84643cb70d5668c0955d5588a7a143a0710b55","observation_id":"b4076495-9561-4ac2-a970-59b3cd42d405","resolution":{"observed_at":"2026-08-04T08:53:05.547904Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-04T08:15:49.310593Z","title":"Reasoning models don't always say what they think","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2510.21978","last_updated":"2026-06-18T16:29:38Z","snapshot_observed_at":"2026-08-09T00:06:37.608642Z","submitted_at":"2025-10-24T19:08:48Z","title":"Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models","version":2},"reference_index":13,"source":"arxiv_source","source_observed_at":"2026-08-04T08:15:49.310593Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2510.21978"},"observation_digest":"sha256:18afed9aef127be7d94ef94ef6c80afc93e507b835297b9c402c9eecf13a89b8","observation_id":"962c22c5-bc25-4353-8cf6-52aa84e33783","resolution":{"observed_at":"2026-08-04T08:15:49.310593Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2511.21931","last_updated":"2025-12-05T21:53:03Z","snapshot_observed_at":"2026-08-11T01:08:08.464571Z","submitted_at":"2025-11-26T21:44:55Z","title":"Does the Model Say What the Data Says? A Simple Heuristic for Model Data Alignment","version":2},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-05-17T04:07:08.752350Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2511.21931"},"observation_digest":"sha256:14092f8381a7bf1d76253400232bbf5923116d5fa04681f459a618cdff0aa2a0","observation_id":"15d4e5a5-a2dc-40f9-b01b-c4b79c422fd2","resolution":{"observed_at":"2026-05-17T04:08:59.918559Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2601.05300","last_updated":"2026-05-04T11:14:47Z","snapshot_observed_at":"2026-08-03T03:42:08.545849Z","submitted_at":"2026-01-08T13:24:49Z","title":"TIME: Temporally Intelligent Meta-reasoning Engine for Context-Triggered Explicit Reasoning","version":2},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-05-16T15:49:26.750021Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2601.05300"},"observation_digest":"sha256:1e229a5dd544f13c864b98a7ad2c2b19776ae14b4ea51dd7c30459e92b795840","observation_id":"e2f94947-bc2e-4e8a-9c83-19f530296e93","resolution":{"observed_at":"2026-05-16T15:51:05.408221Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-03T09:37:06.392506Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2601.13433","last_updated":"2026-05-29T00:45:48Z","snapshot_observed_at":"2026-08-03T09:37:04.326965Z","submitted_at":"2026-01-19T22:37:30Z","title":"Who Endorsed It? Measuring Authority Bias Across Expertise Levels in Language Models","version":4},"reference_index":6,"source":"arxiv_source","source_observed_at":"2026-08-03T09:37:06.392506Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2601.13433"},"observation_digest":"sha256:af173a6a14b4da9801040a0a3bc97383eb58de4f7b5dbbb94631029cd8c29efe","observation_id":"5641d735-1424-4009-adf0-ba2249c06512","resolution":{"observed_at":"2026-08-03T09:37:06.392506Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-02T23:26:29.156561Z","title":"Rea- soning models don’t always say what they think.arXiv preprint arXiv:2505.05410,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2602.13904","last_updated":"2026-07-23T02:27:49Z","snapshot_observed_at":"2026-08-10T06:53:01.769687Z","submitted_at":"2026-02-14T21:53:47Z","title":"Diagnosing Pathological Chain-of-Thought in Reasoning Models","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-02T23:26:29.156561Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2602.13904"},"observation_digest":"sha256:9093e8e9a69a13dcb9f4df6c125102d48efe6dbe4eb933bcbb01f5193399ea51","observation_id":"ee42911f-be32-4e48-ad5d-dd22f84a1634","resolution":{"observed_at":"2026-08-02T23:26:29.156561Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2602.23163","last_updated":"2026-04-29T17:15:03Z","snapshot_observed_at":"2026-07-31T11:33:59.999447Z","submitted_at":"2026-02-26T16:27:24Z","title":"A Decision-Theoretic Formalisation of Steganography With Applications to LLM Monitoring","version":3},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-05-15T19:01:08.076475Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2602.23163"},"observation_digest":"sha256:833274fa454cdaea4747053a142969803c8e4c83f933f7d5779d56bc3ef4b0c9","observation_id":"d052d0bd-f976-4c9f-a265-938c17305c5e","resolution":{"observed_at":"2026-05-15T19:01:29.941429Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2603.16331","last_updated":"2026-05-22T07:55:26Z","snapshot_observed_at":"2026-08-01T19:01:41.446744Z","submitted_at":"2026-03-17T10:03:30Z","title":"Decoding the Critique Mechanism in Large Reasoning Models","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-07-11T11:50:26.030339Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2603.16331"},"observation_digest":"sha256:a72607385c2f938a1caa032b6ef9499b9b797d3ccd8694b40591a469a708ced6","observation_id":"0208175d-d09c-48dc-a403-16ac3708cd52","resolution":{"observed_at":"2026-05-25T06:50:27.509774Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2603.22816","last_updated":"2026-04-13T03:43:20Z","snapshot_observed_at":"2026-07-06T22:50:14.856451Z","submitted_at":"2026-03-24T05:38:13Z","title":"Measuring and curing reasoning rigidity: from decorative chain-of-thought to genuine faithfulness","version":3},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-05-15T01:18:48.798851Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2603.22816"},"observation_digest":"sha256:c502c0249e57caed9f26c1dce7bac72d467dc79bfddea3c77a91610d961022b6","observation_id":"2d8925cd-3703-41b9-b029-08f8412cb9f6","resolution":{"observed_at":"2026-05-15T01:19:37.314423Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.09235","last_updated":"2026-04-10T11:44:27Z","snapshot_observed_at":"2026-07-06T22:58:08.579543Z","submitted_at":"2026-04-10T11:44:27Z","title":"Unreal Thinking: Chain-of-Thought Hijacking via Two-stage Backdoor","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-05-10T17:31:37.034954Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.09235"},"observation_digest":"sha256:0b7f29987e2de79b80bee71a8215e674d50f0dfd62600671f7df5b886e689751","observation_id":"d55b2e55-70c2-4e9f-9061-35dc68a0ca37","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.13602","last_updated":"2026-04-15T08:11:34Z","snapshot_observed_at":"2026-07-06T23:01:32.542040Z","submitted_at":"2026-04-15T08:11:34Z","title":"Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges","version":1},"reference_index":50,"source":"pdf_text","source_observed_at":"2026-05-10T13:58:53.430492Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.13602"},"observation_digest":"sha256:9dbd1e718ea07a435a000e693b604f34e20e17850ff29c1f6c07cfb26a05e781","observation_id":"de4e5c57-26a4-4f22-9482-9b20151a6582","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.14888","last_updated":"2026-08-05T14:02:10Z","snapshot_observed_at":"2026-08-08T23:12:54.979398Z","submitted_at":"2026-04-16T11:28:53Z","title":"Reasoning Dynamics and the Limits of Monitoring Modality Reliance in Vision-Language Models","version":2},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-05-10T10:58:30.370708Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.14888"},"observation_digest":"sha256:359958b1328b7e9d435df6261164f3bf0f992e42f4d4c04a054de1359093a851","observation_id":"1ea296e3-d673-435b-a44f-d0104d2989fa","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.15231","last_updated":"2026-06-01T09:42:47Z","snapshot_observed_at":"2026-07-12T19:46:04.951596Z","submitted_at":"2026-04-16T17:09:30Z","title":"RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-05-10T10:54:35.000783Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.15231"},"observation_digest":"sha256:8628795cacfafa871d78957976506da42c8eace269ed09ab1dd5abf5b0b92cdc","observation_id":"15a8e342-8d97-447d-9c15-1b6d2ef2d903","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-07-12T19:46:06.596310Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2604.15231","last_updated":"2026-06-01T09:42:47Z","snapshot_observed_at":"2026-07-12T19:46:04.951596Z","submitted_at":"2026-04-16T17:09:30Z","title":"RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography","version":2},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-07-12T19:46:06.596310Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.15231"},"observation_digest":"sha256:45fafb8e9da8fd377fe3f7c44a9392ba4a7eb8c1014f94a721b17ddbc0821107","observation_id":"d8fc774a-cf78-4284-9d2d-b5f5eb752d97","resolution":{"observed_at":"2026-07-12T19:46:06.596310Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.15726","last_updated":"2026-04-17T05:59:08Z","snapshot_observed_at":"2026-08-01T01:44:21.919940Z","submitted_at":"2026-04-17T05:59:08Z","title":"LLM Reasoning Is Latent, Not the Chain of Thought","version":1},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-05-10T08:49:05.178087Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.15726"},"observation_digest":"sha256:b8352a15a1a3d0ec64ec7955d4841aa5eba4e9c82cf4bbce28b0033fe7474bec","observation_id":"0e9ff09e-a53d-40a6-90b6-00cb2d7f6d00","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.24178","last_updated":"2026-04-27T08:36:13Z","snapshot_observed_at":"2026-07-06T23:10:16.426836Z","submitted_at":"2026-04-27T08:36:13Z","title":"Meta-Aligner: Bidirectional Preference-Policy Optimization for Multi-Objective LLMs Alignment","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-05-08T04:29:54.013922Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.24178"},"observation_digest":"sha256:55143324d3698d7085bdc7f62dd8074230f8cb13f076e60a85d13edac5bfc724","observation_id":"0ab57bc7-0888-4fce-bbdd-9816e3682432","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.24966","last_updated":"2026-04-27T20:07:09Z","snapshot_observed_at":"2026-07-06T23:10:52.763251Z","submitted_at":"2026-04-27T20:07:09Z","title":"Risk Reporting for Developers' Internal AI Model Use","version":1},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-05-07T17:47:21.321820Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.24966"},"observation_digest":"sha256:da9f2f97a5ac6f135e69187e46829075767a877e7f7a5122944ec56e650be70a","observation_id":"f33251ec-8c90-4958-b480-5098bb9e89d9","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.25110","last_updated":"2026-05-05T18:12:18Z","snapshot_observed_at":"2026-08-06T08:02:49.193188Z","submitted_at":"2026-04-28T01:32:46Z","title":"Knowledge Distillation Must Account for What It Loses","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-05-07T16:58:41.449172Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.25110"},"observation_digest":"sha256:b6a9ef852f888353f4123d2a164a1358c7880d1a94c2bf0ce0c2dfbe26b8497e","observation_id":"f01d4a45-c755-4998-a99a-18d72c0bc82e","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.25110","last_updated":"2026-05-05T18:12:18Z","snapshot_observed_at":"2026-08-06T08:02:49.193188Z","submitted_at":"2026-04-28T01:32:46Z","title":"Knowledge Distillation Must Account for What It Loses","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-05-08T03:31:56.787201Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.25110"},"observation_digest":"sha256:caa3696d04c05a87facc46d33d4605ed3d6384f7142fc81f5061341ae2dc5bd6","observation_id":"8f6e77be-77d3-46cd-8720-222d36b097d2","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.27960","last_updated":"2026-06-11T07:29:25Z","snapshot_observed_at":"2026-07-06T23:13:20.516759Z","submitted_at":"2026-04-30T14:55:48Z","title":"LLMs as ASP Programmers: Self-Correction Enables Task-Agnostic Nonmonotonic Reasoning","version":1},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-05-07T05:51:45.499422Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.27960"},"observation_digest":"sha256:ec843fd5432125e54ea6119a74946bbd4ec3566fdc30f85e613037b229d49f7e","observation_id":"00e2d83c-05f3-4c81-8a42-fd6a2a80bea4","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2604.27960","last_updated":"2026-06-11T07:29:25Z","snapshot_observed_at":"2026-07-06T23:13:20.516759Z","submitted_at":"2026-04-30T14:55:48Z","title":"LLMs as ASP Programmers: Self-Correction Enables Task-Agnostic Nonmonotonic Reasoning","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-07-01T08:07:20.134045Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2604.27960"},"observation_digest":"sha256:d9d587f9d9683f7ac18045e2c7f70f83e121cbc3f9543bd0c2c7126af23e8cf7","observation_id":"4990d3c3-1389-4523-8f49-7d3fbc3c8800","resolution":{"observed_at":"2026-07-01T08:15:31.889929Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.01048","last_updated":"2026-05-01T19:23:33Z","snapshot_observed_at":"2026-07-06T23:14:20.123289Z","submitted_at":"2026-05-01T19:23:33Z","title":"Compared to What? Baselines and Metrics for Counterfactual Prompting","version":1},"reference_index":48,"source":"arxiv_source","source_observed_at":"2026-05-09T19:02:46.991897Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.01048"},"observation_digest":"sha256:3425c246d1cb2c59a864e021dc75c90aa7315c937c7c4152fd294a28be9ac5d3","observation_id":"27acda19-81ed-44f8-aa1c-25997ffef768","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.01164","last_updated":"2026-05-01T23:46:29Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2026-05-01T23:46:29Z","title":"LLMs Should Not Yet Be Credited with Decision Explanation","version":1},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-05-09T18:41:24.552939Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.01164"},"observation_digest":"sha256:a40fd8573e2b04cdea67333e1e812640110434c875d27a6cb2310b8ac9170c48","observation_id":"4d178c74-9cdb-4d7f-b9d5-76aa514d3719","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.06882","last_updated":"2026-05-07T19:31:43Z","snapshot_observed_at":"2026-07-06T23:19:15.382283Z","submitted_at":"2026-05-07T19:31:43Z","title":"How Well Do LLMs Perform on the Simplest Long-Chain Reasoning Tasks: An Empirical Study on the Equivalence Class Problem","version":1},"reference_index":28,"source":"arxiv_source","source_observed_at":"2026-05-11T01:29:33.453354Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.06882"},"observation_digest":"sha256:29afa6f72e74a604ae19693386be995fc67f12d8bb0d7b8c92daf31f46994d13","observation_id":"443bff77-af49-4d53-abb8-cd4e56939818","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.09519","last_updated":"2026-05-10T13:05:44Z","snapshot_observed_at":"2026-08-02T05:03:05.119850Z","submitted_at":"2026-05-10T13:05:44Z","title":"Weighted Rules under the Stable Model Semantics","version":1},"reference_index":54,"source":"arxiv_source","source_observed_at":"2026-05-12T03:38:04.965252Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.09519"},"observation_digest":"sha256:74dd976bb6ccafa7ec43f6fef54e03c85eafdb5824005c827f701cc3635fb4ed","observation_id":"905a0f71-d091-47e3-95df-9c3ae5bb58dc","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.09716","last_updated":"2026-05-10T19:30:16Z","snapshot_observed_at":"2026-08-02T13:42:36.803435Z","submitted_at":"2026-05-10T19:30:16Z","title":"Medical Model Synthesis Architectures: A Case Study","version":1},"reference_index":6,"source":"arxiv_source","source_observed_at":"2026-05-12T03:27:59.466519Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.09716"},"observation_digest":"sha256:300c9d1f1f602a37e3a9552b24727a1f3a30695c5bf75f9cc8165d5dc5d25709","observation_id":"0890147d-6344-46d2-b322-4049a2b5ae88","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.10601","last_updated":"2026-05-11T14:02:56Z","snapshot_observed_at":"2026-07-06T23:22:32.858399Z","submitted_at":"2026-05-11T14:02:56Z","title":"The Open-Box Fallacy: Why AI Deployment Needs a Calibrated Verification Regime","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-05-12T03:37:12.567351Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.10601"},"observation_digest":"sha256:c7ee490c375dd051cae86fe02f3a08d20d79ea5ec9b445f09a175ef6870db372","observation_id":"1379bc7a-10d2-4f7c-bb5c-e48a32918246","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.10930","last_updated":"2026-05-15T20:34:22Z","snapshot_observed_at":"2026-07-06T23:22:52.369277Z","submitted_at":"2026-05-11T17:58:12Z","title":"Evaluating the False Trust Engendered by LLM Explanations","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-05-12T03:27:29.025166Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.10930"},"observation_digest":"sha256:86df0706543b9e18ca80f5acb2aa50296aef712b82dfa68f3e5bd99b77b82e2b","observation_id":"97761f00-d0d9-4985-8d59-97100c0a2408","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.10930","last_updated":"2026-05-15T20:34:22Z","snapshot_observed_at":"2026-07-06T23:22:52.369277Z","submitted_at":"2026-05-11T17:58:12Z","title":"Evaluating the False Trust Engendered by LLM Explanations","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-05-20T21:53:37.211145Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.10930"},"observation_digest":"sha256:96f1995f1cf06996c358828ba654734f40801d39aea149451b0fe6e107e4a475","observation_id":"18fac225-07d6-4ccb-9680-e9ea6f8dc4c4","resolution":{"observed_at":"2026-05-20T21:53:47.150254Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.11467","last_updated":"2026-05-12T03:30:23Z","snapshot_observed_at":"2026-07-06T23:23:16.461539Z","submitted_at":"2026-05-12T03:30:23Z","title":"Drop the Act: Probe-Filtered RL for Faithful Chain-of-Thought Reasoning","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-05-13T02:07:59.111403Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.11467"},"observation_digest":"sha256:c6475565e51f62429c1f098260d4d5795d9c537d533982122127782a6f43df6d","observation_id":"a6c9611b-04e9-46f6-a459-6342b5d71e30","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.11746","last_updated":"2026-05-12T08:24:47Z","snapshot_observed_at":"2026-08-03T19:32:09.592136Z","submitted_at":"2026-05-12T08:24:47Z","title":"When Reasoning Traces Become Performative: Step-Level Evidence that Chain-of-Thought Is an Imperfect Oversight Channel","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-05-13T06:30:12.558660Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.11746"},"observation_digest":"sha256:3912cd5cb82ff2cd0922b914f7d4c15cffa9b4060573f3201d241d1fe9e56731","observation_id":"298e183b-3d79-4f02-82bb-6a5e067eaa22","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-07-15T11:13:18.391956Z","title":"Advances in Neural Information Processing Systems, 37:45767–45790","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2605.12459","last_updated":"2026-07-14T15:12:16Z","snapshot_observed_at":"2026-07-17T23:19:03.090481Z","submitted_at":"2026-05-12T17:45:56Z","title":"Trace ideals and uniserial modules","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-07-15T11:13:18.391956Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.12459"},"observation_digest":"sha256:6c6b079e67614904de01b8773be4fba1bab051510a7149ce73fe510ce34fb596","observation_id":"cd56e294-43cc-4ae3-a435-b5a8a2df0670","resolution":{"observed_at":"2026-07-15T11:13:18.391956Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.12460","last_updated":"2026-05-12T17:47:41Z","snapshot_observed_at":"2026-08-02T23:57:53.111604Z","submitted_at":"2026-05-12T17:47:41Z","title":"Multi-Stream LLMs: Unblocking Language Models with Parallel Streams of Thoughts, Inputs and Outputs","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-05-13T05:56:11.373242Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.12460"},"observation_digest":"sha256:1306d19c890e7d15efc8ebcd39b52503604d75fc131cc9eb12450cc7af8e6914","observation_id":"2d5bbaf8-b8ad-4afa-8f43-8315ffe1a154","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.12673","last_updated":"2026-05-12T19:22:45Z","snapshot_observed_at":"2026-08-02T02:31:32.394072Z","submitted_at":"2026-05-12T19:22:45Z","title":"Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack","version":1},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-05-14T20:31:50.043920Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.12673"},"observation_digest":"sha256:7fc663aed1d6e5f93623e02c03dc45115177625c4431f0f0b331e3469624e4b0","observation_id":"24b201f8-68b0-46b3-aafc-03a13b094649","resolution":{"observed_at":"2026-05-14T20:32:56.763151Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.12746","last_updated":"2026-05-12T20:49:30Z","snapshot_observed_at":"2026-07-06T23:24:23.402304Z","submitted_at":"2026-05-12T20:49:30Z","title":"CoT-Guard: Small Models for Strong Monitoring","version":1},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-05-14T19:39:16.264272Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.12746"},"observation_digest":"sha256:b7d52ed1f02530a1433e779cd374f56e7a3648b9756294b71abb75812e1fee2e","observation_id":"7df7c8d5-f52b-4b61-afeb-25c3fea8e00a","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.13290","last_updated":"2026-05-13T10:04:38Z","snapshot_observed_at":"2026-08-03T18:35:13.170763Z","submitted_at":"2026-05-13T10:04:38Z","title":"What properties of reasoning supervision are associated with improved downstream model quality?","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-05-14T19:36:30.699633Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.13290"},"observation_digest":"sha256:adbef9a2ac0dcb7f7c1002b97fa74af6a4f36988a10e956271e0a5835a963ad1","observation_id":"14716553-4a52-4687-a8cb-4b99e8b0d481","resolution":{"observed_at":"2026-05-14T20:18:01.027228Z","resolver_source":"arxiv_id","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.14049","last_updated":"2026-05-13T19:11:09Z","snapshot_observed_at":"2026-08-02T04:57:00.664517Z","submitted_at":"2026-05-13T19:11:09Z","title":"Bridging Legal Interpretation and Formal Logic: Faithfulness, Assumption, and the Future of AI Legal Reasoning","version":1},"reference_index":19,"source":"arxiv_source","source_observed_at":"2026-05-15T05:23:11.698521Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.14049"},"observation_digest":"sha256:714f1f5dcb0163a49776876d884247ca63eeb67a7d8711e28168118b8aec122c","observation_id":"72267283-16c0-42eb-a0c2-f9d345092281","resolution":{"observed_at":"2026-05-15T05:25:04.091839Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.14415","last_updated":"2026-05-14T06:04:40Z","snapshot_observed_at":"2026-07-06T23:25:49.545418Z","submitted_at":"2026-05-14T06:04:40Z","title":"SWE-Chain: Benchmarking Coding Agents on Chained Release-Level Package Upgrades","version":1},"reference_index":86,"source":"arxiv_source","source_observed_at":"2026-05-15T02:31:18.183715Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.14415"},"observation_digest":"sha256:adc2664774f2929c9bfa08af71b6f9f3bb2c464559231fe6db639e9894bf145c","observation_id":"0c687c55-bc96-45e7-bec1-1d32ecf88a00","resolution":{"observed_at":"2026-05-15T02:33:32.331296Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.16538","last_updated":"2026-05-15T18:33:46Z","snapshot_observed_at":"2026-07-06T23:27:38.955917Z","submitted_at":"2026-05-15T18:33:46Z","title":"LLMs in Qualitative Research: Opportunities, Limitations, and Practical Considerations","version":1},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-05-20T15:59:56.231210Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.16538"},"observation_digest":"sha256:aa1feb1557248288d1bad3ad92c10af2e9951d9cc7bb4ee44cdffdcc08eef59b","observation_id":"2bdf5cbd-10a0-41f5-aab4-9f883173e6a3","resolution":{"observed_at":"2026-05-20T16:03:33.155685Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.18549","last_updated":"2026-05-18T15:29:04Z","snapshot_observed_at":"2026-08-03T06:08:37.339767Z","submitted_at":"2026-05-18T15:29:04Z","title":"Monitoring the Internal Monologue: Probe Trajectories Reveal Reasoning Dynamics","version":1},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-05-20T10:54:20.285841Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.18549"},"observation_digest":"sha256:cd5ec629cb661179d9d6deddf81474691062c0bb3f40290d26b0cae1e5b2bf70","observation_id":"8b6033d8-dc92-4d3b-a1ed-04c886432846","resolution":{"observed_at":"2026-05-20T11:03:13.933854Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.19943","last_updated":"2026-05-19T15:00:38Z","snapshot_observed_at":"2026-08-06T08:53:20.073219Z","submitted_at":"2026-05-19T15:00:38Z","title":"Probabilistic Tiny Recursive Model","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-05-20T05:39:38.909349Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.19943"},"observation_digest":"sha256:0158410f659877df1a64dba5ce34b5804d97202333c3514d8cf6091da46007d7","observation_id":"5de14b6c-8e1f-4937-b5c3-9ee0e0899c22","resolution":{"observed_at":"2026-05-20T05:43:05.907422Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.20098","last_updated":"2026-06-09T11:06:35Z","snapshot_observed_at":"2026-08-03T00:08:59.222529Z","submitted_at":"2026-05-19T16:49:29Z","title":"Neurosymbolic Learning for Inference-Time Argumentation","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-05-20T05:11:44.118384Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.20098"},"observation_digest":"sha256:94fe2a379318483ab9a2749883b9f944c064796013d38cce9befe78d5c55adbe","observation_id":"b0f81a3c-ea24-4265-ba64-297fdec4f1e5","resolution":{"observed_at":"2026-05-20T05:13:03.466611Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.20098","last_updated":"2026-06-09T11:06:35Z","snapshot_observed_at":"2026-08-03T00:08:59.222529Z","submitted_at":"2026-05-19T16:49:29Z","title":"Neurosymbolic Learning for Inference-Time Argumentation","version":2},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-06-30T18:04:21.013167Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.20098"},"observation_digest":"sha256:b98fee68da04c31c212165a7db0be4f71502b201d4f497bd6d1d79607a201a6f","observation_id":"3338abe3-aff3-4a91-9965-6d1cae5ca940","resolution":{"observed_at":"2026-06-30T18:04:57.778119Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.21260","last_updated":"2026-05-20T14:51:20Z","snapshot_observed_at":"2026-07-29T20:35:26.564819Z","submitted_at":"2026-05-20T14:51:20Z","title":"On the Cost and Benefit of Chain of Thought: A Learning-Theoretic Perspective","version":1},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-05-21T05:09:37.588841Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.21260"},"observation_digest":"sha256:fe66cd9043e5827e32314d6e0e0e857c2fb89750f151f5c3201f2bb0cbd81059","observation_id":"72b1b6c2-5744-47eb-b542-7568fe0aa45f","resolution":{"observed_at":"2026-05-21T05:13:58.676304Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.22870","last_updated":"2026-05-20T00:32:41Z","snapshot_observed_at":"2026-08-02T01:45:01.340059Z","submitted_at":"2026-05-20T00:32:41Z","title":"The Readout Shortcut: Positional Number Copying Dominates Arithmetic CoT Readout in Small Language Models","version":1},"reference_index":4,"source":"arxiv_source","source_observed_at":"2026-05-25T06:27:42.024094Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.22870"},"observation_digest":"sha256:960a6134b143da0498205ce3888f0d34b2c2847a7dfad29ab97b1b816d5b17ec","observation_id":"036aa5de-343c-4c35-a9cb-dba17059b446","resolution":{"observed_at":"2026-05-25T06:30:25.628182Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.24396","last_updated":"2026-08-10T01:19:49Z","snapshot_observed_at":"2026-08-11T03:21:07.675744Z","submitted_at":"2026-05-23T04:42:45Z","title":"Understanding and Mitigating Premature Confidence for Better LLM Reasoning","version":1},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-06-30T14:03:25.913615Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.24396"},"observation_digest":"sha256:f89fafb70351665778cf18886120d6d2786c68d206a1fd52af841ca18b27e96d","observation_id":"93f60722-4667-4c28-bae8-603703863689","resolution":{"observed_at":"2026-06-30T14:04:44.433322Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.24960","last_updated":"2026-05-24T09:16:55Z","snapshot_observed_at":"2026-07-06T23:34:57.148983Z","submitted_at":"2026-05-24T09:16:55Z","title":"Investigating the Interplay between Contextual and Parametric Chain-of-Thought Faithfulness under Optimization","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-06-30T12:17:12.602012Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.24960"},"observation_digest":"sha256:8fc4f5cf790946cc3ce143dc2022ecf0d86156d817fe91f79e8a25ee06dae96f","observation_id":"36615e4a-62d5-42f8-a29c-05719f7a5fe3","resolution":{"observed_at":"2026-06-30T12:24:39.911749Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.25052","last_updated":"2026-05-24T12:57:01Z","snapshot_observed_at":"2026-08-10T22:06:13.844554Z","submitted_at":"2026-05-24T12:57:01Z","title":"Faithfulness Metrics Don't Measure Faithfulness: A Meta-Evaluation with Ground Truth","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-06-30T11:56:53.355299Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.25052"},"observation_digest":"sha256:8ed5921bb0c7b2aa0f400c9a28c00c10b7dc22655da308832dd803301061903e","observation_id":"1267243d-b84a-4a28-a165-467128bd3df1","resolution":{"observed_at":"2026-06-30T12:04:39.043070Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2605.29087","last_updated":"2026-05-27T20:41:08Z","snapshot_observed_at":"2026-08-08T01:47:50.125564Z","submitted_at":"2026-05-27T20:41:08Z","title":"The Chain Holds, the Answer Folds: Trace-Answer Dissociation in Reasoning Models Under Adversarial Pressure","version":1},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-06-29T11:56:54.670979Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2605.29087"},"observation_digest":"sha256:4d7281aef6b5e251faaa2748e6452fd6eed00c4729061976b71d7e1d1029074a","observation_id":"cc117f66-49fd-45d4-9e01-d012482f881e","resolution":{"observed_at":"2026-06-29T12:03:24.170793Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.00497","last_updated":"2026-05-30T03:00:28Z","snapshot_observed_at":"2026-07-31T17:49:49.063236Z","submitted_at":"2026-05-30T03:00:28Z","title":"\"I Strongly Suspect This Website Is a Scam\": Benchmarking PII Leakage and Detection without Defense in Autonomous Web Agents","version":1},"reference_index":118,"source":"arxiv_source","source_observed_at":"2026-06-28T18:53:41.420255Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.00497"},"observation_digest":"sha256:0b278982ab63f37aa7e1f7782e3302ff98020ced27c49c0fa0c98a6138f62a93","observation_id":"ea7361b8-2675-4d0b-b8d1-99997e495135","resolution":{"observed_at":"2026-06-28T19:42:36.153240Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.00642","last_updated":"2026-05-30T09:37:55Z","snapshot_observed_at":"2026-08-04T11:38:58.569413Z","submitted_at":"2026-05-30T09:37:55Z","title":"Hidden Thoughts Are Not Secret: Reasoning Trace Exposure in LLMs","version":1},"reference_index":46,"source":"arxiv_source","source_observed_at":"2026-06-28T18:48:36.147561Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.00642"},"observation_digest":"sha256:6435d93ab1c70a25d511edf5f7efb5becdddb3090a01d8a9e2ae9de7e6bfaae0","observation_id":"5a3e24ab-bec8-4d1b-b8b5-2164b985e945","resolution":{"observed_at":"2026-06-28T19:52:35.744075Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.03969","last_updated":"2026-06-02T17:53:45Z","snapshot_observed_at":"2026-08-08T07:43:20.537989Z","submitted_at":"2026-06-02T17:53:45Z","title":"Quantifying Faithful Confidence Expression in Large Reasoning Models","version":1},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-06-28T10:24:38.335417Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.03969"},"observation_digest":"sha256:35f0896975d6c171f04bf68bcf4babb137b2d439f59d1429d8d51729174fe571","observation_id":"14c8dcfc-e9df-4fe1-b08c-47d49a33643b","resolution":{"observed_at":"2026-07-02T03:06:29.237014Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.05567","last_updated":"2026-06-04T01:28:36Z","snapshot_observed_at":"2026-08-08T04:18:17.880029Z","submitted_at":"2026-06-04T01:28:36Z","title":"ZERO-APT: A Closed-Loop Adversarial Framework for LLM-Driven Automated Penetration Testing under Intelligent Defense","version":1},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-06-28T01:26:22.434043Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.05567"},"observation_digest":"sha256:57b18794f92d8aa4870fd392e8438b8bab33ca5373f3311186575e7c44936af8","observation_id":"0edb1f89-ee21-4a21-811b-fe05be39555c","resolution":{"observed_at":"2026-07-02T13:16:58.775410Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.05976","last_updated":"2026-07-31T08:26:14Z","snapshot_observed_at":"2026-08-05T23:10:45.588490Z","submitted_at":"2026-06-04T10:17:00Z","title":"The Self-Correction Illusion: Role Relabeling Gates Explicit Error Flagging in Large Language Models","version":1},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-06-28T01:25:07.890796Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.05976"},"observation_digest":"sha256:d2e1b25aefb4f9e8f14f21671193b80f71dd7022d86a71d9978e15b919423a2d","observation_id":"abf2824c-88dc-4bea-8a86-df21dcbead16","resolution":{"observed_at":"2026-06-28T01:31:29.339523Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.06223","last_updated":"2026-07-15T02:59:20Z","snapshot_observed_at":"2026-08-08T18:26:46.307588Z","submitted_at":"2026-06-04T14:34:31Z","title":"From Reward-Hack Activations to Agentic Risk States: Context-Calibrated Mechanistic Monitoring in LLM Agents","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-06-28T01:07:17.014002Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.06223"},"observation_digest":"sha256:00f39c6de53731313dc67e6176f1caebaaacc327f861a5300bddeac1c2176a47","observation_id":"b22e2154-4f51-47c1-a388-6dbf500856e1","resolution":{"observed_at":"2026-07-02T13:36:59.388654Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-02T12:19:46.889450Z","title":"Gao, L., la Tour, T","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2606.06223","last_updated":"2026-07-15T02:59:20Z","snapshot_observed_at":"2026-08-08T18:26:46.307588Z","submitted_at":"2026-06-04T14:34:31Z","title":"From Reward-Hack Activations to Agentic Risk States: Context-Calibrated Mechanistic Monitoring in LLM Agents","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-02T12:19:46.889450Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.06223"},"observation_digest":"sha256:6c1109a9a542c115d49b26c6ff5c551cc164f4d3a7ed1295d0877cd5d7aafe45","observation_id":"9efaf398-c88f-4b7f-aaff-b12045c5eb38","resolution":{"observed_at":"2026-08-02T12:19:46.889450Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.08629","last_updated":"2026-06-07T13:47:11Z","snapshot_observed_at":"2026-08-07T04:03:10.261869Z","submitted_at":"2026-06-07T13:47:11Z","title":"Sycophancy Towards Researchers Drives Performative Misalignment","version":1},"reference_index":63,"source":"arxiv_source","source_observed_at":"2026-06-27T18:52:29.375027Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.08629"},"observation_digest":"sha256:1154acc10cd4c13146dae7b0e843f28624f436ff204229b3311ade04fc2a211a","observation_id":"ced19d64-699c-42a9-9dfb-ba8fc5a9e584","resolution":{"observed_at":"2026-07-02T22:27:26.205760Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.09711","last_updated":"2026-06-08T16:32:54Z","snapshot_observed_at":"2026-07-06T23:49:03.237958Z","submitted_at":"2026-06-08T16:32:54Z","title":"Proxy Reward Internalization and Mechanistic Exploitation: A Learned Precursor to Reward Hacking and Its Generalization","version":1},"reference_index":239,"source":"arxiv_source","source_observed_at":"2026-06-27T16:26:34.918099Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.09711"},"observation_digest":"sha256:622fe6f7460cfeaee18bf18ec235d82ad0a1f60f819ee87d2fc98ebaec6967e8","observation_id":"a6ef89e3-3b6c-49e2-9372-19cded203eeb","resolution":{"observed_at":"2026-07-03T01:37:30.562997Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.10456","last_updated":"2026-06-09T06:04:30Z","snapshot_observed_at":"2026-08-07T22:43:09.547652Z","submitted_at":"2026-06-09T06:04:30Z","title":"The Distributed Detectability Band Against Marginal-Preserving Attacks","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-06-27T12:58:22.056355Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.10456"},"observation_digest":"sha256:007ec835681964bb375d162c68ea3831327abbf3821feaf7925e601d4d70da9d","observation_id":"ed797390-a567-404b-af25-f504bd1c365f","resolution":{"observed_at":"2026-07-03T05:57:41.849792Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.12689","last_updated":"2026-06-10T21:23:22Z","snapshot_observed_at":"2026-07-06T23:51:35.904251Z","submitted_at":"2026-06-10T21:23:22Z","title":"Observable Patterns Are Not Explanations: A Causal-Geometric Analysis of Latent Reasoning Models","version":1},"reference_index":30,"source":"arxiv_source","source_observed_at":"2026-06-27T09:36:05.700067Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.12689"},"observation_digest":"sha256:c00f9d9e18bbd8adbd6765c36954d6eb8bb3ffdcb8c5b270570e655ced19d742","observation_id":"f4fc79be-0cac-4a79-9e02-5734febcfbbf","resolution":{"observed_at":"2026-07-03T11:18:03.796570Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.15127","last_updated":"2026-06-19T06:10:08Z","snapshot_observed_at":"2026-07-06T23:52:32.998232Z","submitted_at":"2026-06-13T05:41:57Z","title":"Beyond Accuracy: Measuring Bias Acknowledgment in Chain-of-Thought Reasoning for Responsible AI Evaluation","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-06-27T04:43:04.740925Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.15127"},"observation_digest":"sha256:a995b319f4e70d4a7a121e7a0d555a80990676b019a0b94bab5d7ee7cfcbc375","observation_id":"60250f88-3172-43ba-bf75-dd6341c2c6f1","resolution":{"observed_at":"2026-07-03T16:58:43.404342Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.16118","last_updated":"2026-06-19T01:41:59Z","snapshot_observed_at":"2026-07-06T23:52:37.922109Z","submitted_at":"2026-06-15T02:14:49Z","title":"Know Your Limits : On the Faithfulness of LLMs as Solvers and Autoformalizers in Legal Reasoning","version":2},"reference_index":3,"source":"arxiv_source","source_observed_at":"2026-06-27T04:24:11.949884Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.16118"},"observation_digest":"sha256:ec4ef4507043356cb52b582e6aa9da20e3adb646f73cdae2c06e9b380b588061","observation_id":"8e9fc536-173e-4202-ad20-be7a769556ad","resolution":{"observed_at":"2026-07-03T17:18:43.460462Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.19588","last_updated":"2026-06-17T20:44:03Z","snapshot_observed_at":"2026-07-06T23:54:51.451358Z","submitted_at":"2026-06-17T20:44:03Z","title":"Analyzing the Narration Gap in LLM-Solver Loops","version":1},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-06-26T20:41:54.964459Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.19588"},"observation_digest":"sha256:dacd9a453216bd8c7c004e5b331a238192fc3e5d13edc27adffee8f113849743","observation_id":"166f1bd7-68e5-493a-8294-b0190d2ace62","resolution":{"observed_at":"2026-06-26T20:49:57.677402Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.21821","last_updated":"2026-06-20T01:18:58Z","snapshot_observed_at":"2026-08-08T12:53:22.790848Z","submitted_at":"2026-06-20T01:18:58Z","title":"Local Causal Attribution of Chain-of-Thought Reasoning","version":1},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-07-11T11:50:26.030339Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.21821"},"observation_digest":"sha256:413693c36f9686bb26cb3be9162d8851069d930ee4ba09802a40f8829bb440ff","observation_id":"e3156763-063a-496d-a562-cc7412994ac9","resolution":{"observed_at":"2026-06-26T12:49:29.099013Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.23597","last_updated":"2026-06-22T17:02:53Z","snapshot_observed_at":"2026-08-04T13:40:03.543286Z","submitted_at":"2026-06-22T17:02:53Z","title":"Against Proxy Optimization","version":1},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-06-26T08:26:50.595370Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.23597"},"observation_digest":"sha256:b058428ffe91574642174126417b74fcbcc824a4e8569ca24deed31f13208827","observation_id":"1cc15de3-a976-4434-bcec-2e1b23359500","resolution":{"observed_at":"2026-07-04T10:49:46.634634Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.28358","last_updated":"2026-06-09T07:06:13Z","snapshot_observed_at":"2026-07-07T00:02:29.000757Z","submitted_at":"2026-06-09T07:06:13Z","title":"How Do LLMs Cite? A Mechanistic Interpretation of Attribution in Retrieval-Augmented Generation","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-06-30T11:12:02.227976Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.28358"},"observation_digest":"sha256:b424e1577ee9fce4a9094e8ebe4d513deb3a1f184858e7910fa3cc95be8f1629","observation_id":"1023fca4-0c2c-4ba5-bf14-f47e8ba7b344","resolution":{"observed_at":"2026-06-30T11:14:36.929953Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.28615","last_updated":"2026-06-26T21:14:56Z","snapshot_observed_at":"2026-08-04T19:40:32.344244Z","submitted_at":"2026-06-26T21:14:56Z","title":"What LLMs explain is not what they believe: Evaluating explanation sufficiency under models' own input beliefs","version":1},"reference_index":133,"source":"arxiv_source","source_observed_at":"2026-06-30T00:38:21.949283Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.28615"},"observation_digest":"sha256:f1160d184096df5ad90055195847796f44b99f34741ebad681d9ed3e712f70e9","observation_id":"19ccb1b6-dec5-480c-a109-8a5ce6b01850","resolution":{"observed_at":"2026-07-01T16:25:50.038326Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.28863","last_updated":"2026-06-27T11:12:17Z","snapshot_observed_at":"2026-07-07T00:02:52.539331Z","submitted_at":"2026-06-27T11:12:17Z","title":"Defeat Devices in AI Systems","version":1},"reference_index":12,"source":"arxiv_source","source_observed_at":"2026-06-30T08:34:58.879346Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.28863"},"observation_digest":"sha256:c5f946b5c695a50506a16cff3c2caf0cd0ae494d2c193928f3fadc89c24dba73","observation_id":"bea6fe0e-0506-4155-ba9b-1b29eb4c9774","resolution":{"observed_at":"2026-06-30T08:44:27.899211Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2606.32038","last_updated":"2026-06-30T17:59:32Z","snapshot_observed_at":"2026-08-02T04:33:32.900362Z","submitted_at":"2026-06-30T17:59:32Z","title":"Introspective Coupling: Self-Explanation Training Tracks Behavioral Change Despite Fixed Supervision","version":1},"reference_index":5,"source":"arxiv_source","source_observed_at":"2026-07-01T05:18:30.190239Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2606.32038"},"observation_digest":"sha256:a3dc8ff8ec2348fb4ce80b3c4b49057d64c5fefa6a9213954a51fbde399016c2","observation_id":"cf4e6467-6a82-45aa-a571-d41bfbbb3270","resolution":{"observed_at":"2026-07-01T10:35:42.816633Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2607.01047","last_updated":"2026-07-01T15:08:02Z","snapshot_observed_at":"2026-08-07T14:53:02.059078Z","submitted_at":"2026-07-01T15:08:02Z","title":"Conversable Complexity: Agentic LLM Collectives as Interpretable Substrates","version":1},"reference_index":82,"source":"arxiv_source","source_observed_at":"2026-07-02T12:46:40.700728Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2607.01047"},"observation_digest":"sha256:0387ae94ac3565fe128e5f99d064e2c2f9944cf0fa672b1a703c2d9327f55799","observation_id":"3c96079a-738b-489f-9f4e-b7300b8b2eb0","resolution":{"observed_at":"2026-07-02T12:46:56.222938Z","resolver_source":"local_arxiv","status":"metadata_mismatch"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-07-12T02:02:49.492856Z","title":"arXiv preprint arXiv:2505.05410 , year=","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2607.03502","last_updated":"2026-07-03T17:18:34Z","snapshot_observed_at":"2026-08-03T03:53:19.375179Z","submitted_at":"2026-07-03T17:18:34Z","title":"Reading Between the Dots: Decoding Hidden Computation across Filler Tokens","version":1},"reference_index":13,"source":"arxiv_source","source_observed_at":"2026-07-12T02:02:49.492856Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2607.03502"},"observation_digest":"sha256:9c35be9a26c0e782a144b9b38e4795422956408554fb53d37df5f2f6870a793d","observation_id":"632a689a-da3f-445c-bf59-005be81801ac","resolution":{"observed_at":"2026-07-12T02:02:49.492856Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":"2505.05410","doi":"10.48550/arxiv.2210.14889","metadata_source":"pith","pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-05T02:28:24.338817Z","title":"Reasoning Models Don't Always Say What They Think","venue":"cs.CL","work_id":"b9bdcbf5-9ae0-464c-b1a6-de04f85a6e33","year":2025},"citing_paper":{"arxiv_id":"2607.08173","last_updated":"2026-07-09T07:19:05Z","snapshot_observed_at":"2026-08-03T05:20:57.555833Z","submitted_at":"2026-07-09T07:19:05Z","title":"Overthinking: Amplifying Reasoning Weights to Extract Learned Secrets","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-07-10T11:54:32.051780Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2607.08173"},"observation_digest":"sha256:70db441bcc2ab68440ade6d5a55149f5b42785750f6106c3071b6e3b5503e490","observation_id":"80f741cc-f051-4447-8d5a-6a40f24b9ea1","resolution":{"observed_at":"2026-07-10T11:57:03.259382Z","resolver_source":"local_arxiv","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-10T06:31:04.303077+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-07-14T15:45:54.532529Z","title":"Bowman, Jan Leike, Jared Kaplan, and Ethan Perez","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.09786","last_updated":"2026-08-02T18:21:10Z","snapshot_observed_at":"2026-08-07T05:12:04.880594Z","submitted_at":"2026-07-08T14:18:26Z","title":"Length Penalties Make Chain-of-Thought Less Monitorable","version":1},"reference_index":52,"source":"arxiv_source","source_observed_at":"2026-07-14T15:45:54.532529Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2607.09786"},"observation_digest":"sha256:562144a12e36606033a33b8132aad211ee056e1a956061543bfad1bc63e2bd70","observation_id":"f7628fda-e636-4363-ae46-64d25abe2a80","resolution":{"observed_at":"2026-07-14T15:45:54.532529Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05410","snapshot_observed_at":"2026-08-02T08:06:10.773565Z","title":"Bowman, Jan Leike, Jared Kaplan, and Ethan Perez","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.09786","last_updated":"2026-08-02T18:21:10Z","snapshot_observed_at":"2026-08-07T05:12:04.880594Z","submitted_at":"2026-07-08T14:18:26Z","title":"Length Penalties Make Chain-of-Thought Less Monitorable","version":2},"reference_index":53,"source":"arxiv_source","source_observed_at":"2026-08-02T08:06:10.773565Z"},"links":{"cited_paper":"/paper/2505.05410","citing_paper":"/paper/2607.09786"},"observation_digest":"sha256:7e976eabf973931ab4b760bf35fa423fe12a056b55de4eb3dfea91e5b5745a34","observation_id":"f25cadd4-831e-4196-aafc-d85e71457bf1","resolution":{"observed_at":"2026-08-02T08:06:10.773565Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"links":{"evidence":"/evidence","html":"/paper/2505.05410/citation-record","integrity":"/paper/2505.05410/integrity","json":"/paper/2505.05410/citation-record.json","paper":"/paper/2505.05410"},"outbound":[],"paper":{"arxiv_id":"2505.05410","last_updated":"2025-05-08T16:51:43Z","latest_version":1,"primary_category":"cs.CL","snapshot_observed_at":"2026-08-11T01:19:00.494448Z","submitted_at":"2025-05-08T16:51:43Z","title":"Reasoning Models Don't Always Say What They Think"},"reference_resolution":{"displayed":0,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":0,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":0},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-10T06:31:04.303077+00:00","source":"crossref"},{"observed_at":"2026-08-10T06:30:57.382061+00:00","source":"retraction_watch"}],"thesis":"As of 11 August 2026, this Paper Citation Record lists 0 of 0 outbound references and 100 inbound Pith citation observations for arXiv:2505.05410."}