{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:3GRGBG2MBRAVMQCPHH57NZI2QA","short_pith_number":"pith:3GRGBG2M","canonical_record":{"source":{"id":"2310.18512","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T22:02:29Z","cross_cats_sorted":[],"title_canon_sha256":"59d0c8f3daf91c66baa3c2201d83d63d2a9f62e52b9eb1d23d07f90ea88f8810","abstract_canon_sha256":"cb6a9831e895e3a02b78daa9ae61d223a1f19c6749bcd5dfd6be37ddc94dcb1f"},"schema_version":"1.0"},"canonical_sha256":"d9a2609b4c0c4156404f39fbf6e51a800819df044ef4b5d661fa6f17cfc79102","source":{"kind":"arxiv","id":"2310.18512","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.18512","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"arxiv_version","alias_value":"2310.18512v2","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.18512","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"pith_short_12","alias_value":"3GRGBG2MBRAV","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"pith_short_16","alias_value":"3GRGBG2MBRAVMQCP","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"pith_short_8","alias_value":"3GRGBG2M","created_at":"2026-07-05T07:07:40Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:3GRGBG2MBRAVMQCPHH57NZI2QA","target":"record","payload":{"canonical_record":{"source":{"id":"2310.18512","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T22:02:29Z","cross_cats_sorted":[],"title_canon_sha256":"59d0c8f3daf91c66baa3c2201d83d63d2a9f62e52b9eb1d23d07f90ea88f8810","abstract_canon_sha256":"cb6a9831e895e3a02b78daa9ae61d223a1f19c6749bcd5dfd6be37ddc94dcb1f"},"schema_version":"1.0"},"canonical_sha256":"d9a2609b4c0c4156404f39fbf6e51a800819df044ef4b5d661fa6f17cfc79102","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:07:40.290159Z","signature_b64":"xmEij8+zXAIH51zdRGmaZKwc1yDqGPOyZm8Xum9/x27aAG9YMofpaTUW3H0JaMEh3Ot1OIqYyWzAWxfccbJHCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d9a2609b4c0c4156404f39fbf6e51a800819df044ef4b5d661fa6f17cfc79102","last_reissued_at":"2026-07-05T07:07:40.289697Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:07:40.289697Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2310.18512","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:07:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fW6ZSCfv0DoyFiIUeq6qjLimsg6/Y5CCbUSlt0DTK7Xs0FmVLHHF5Z62wY0xzdG5T3IR+UWGGZhLoMP8N1W1CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T03:10:23.306584Z"},"content_sha256":"247a3aee97a5b2e9bcee3b8453992c834ee0cafd243c469a15416c42f070f7e8","schema_version":"1.0","event_id":"sha256:247a3aee97a5b2e9bcee3b8453992c834ee0cafd243c469a15416c42f070f7e8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:3GRGBG2MBRAVMQCPHH57NZI2QA","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Preventing Language Models From Hiding Their Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Fabien Roger, Ryan Greenblatt","submitted_at":"2023-10-27T22:02:29Z","abstract_excerpt":"Large language models (LLMs) often benefit from intermediate steps of reasoning to generate answers to complex problems. When these intermediate steps of reasoning are used to monitor the activity of the model, it is essential that this explicit reasoning is faithful, i.e. that it reflects what the model is actually reasoning about. In this work, we focus on one potential way intermediate steps of reasoning could be unfaithful: encoded reasoning, where an LLM could encode intermediate steps of reasoning in the generated text in a way that is not understandable to human readers. We show that la"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.18512","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2310.18512/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:07:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"aeQwnFShqnMAWjkISZR35OOq7lTaatioLx+qQEcX80AGUWrklwBBnYySxjH0veSGuTceU4E25z/YnUPsNluPBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T03:10:23.307179Z"},"content_sha256":"ba4761e8b5da1334f71ea1b2abed08f97c36328f0f52c0e52bf120d22fc73102","schema_version":"1.0","event_id":"sha256:ba4761e8b5da1334f71ea1b2abed08f97c36328f0f52c0e52bf120d22fc73102"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3GRGBG2MBRAVMQCPHH57NZI2QA/bundle.json","state_url":"https://pith.science/pith/3GRGBG2MBRAVMQCPHH57NZI2QA/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3GRGBG2MBRAVMQCPHH57NZI2QA/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-06T03:10:23Z","links":{"resolver":"https://pith.science/pith/3GRGBG2MBRAVMQCPHH57NZI2QA","bundle":"https://pith.science/pith/3GRGBG2MBRAVMQCPHH57NZI2QA/bundle.json","state":"https://pith.science/pith/3GRGBG2MBRAVMQCPHH57NZI2QA/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3GRGBG2MBRAVMQCPHH57NZI2QA/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:3GRGBG2MBRAVMQCPHH57NZI2QA","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"cb6a9831e895e3a02b78daa9ae61d223a1f19c6749bcd5dfd6be37ddc94dcb1f","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T22:02:29Z","title_canon_sha256":"59d0c8f3daf91c66baa3c2201d83d63d2a9f62e52b9eb1d23d07f90ea88f8810"},"schema_version":"1.0","source":{"id":"2310.18512","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.18512","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"arxiv_version","alias_value":"2310.18512v2","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.18512","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"pith_short_12","alias_value":"3GRGBG2MBRAV","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"pith_short_16","alias_value":"3GRGBG2MBRAVMQCP","created_at":"2026-07-05T07:07:40Z"},{"alias_kind":"pith_short_8","alias_value":"3GRGBG2M","created_at":"2026-07-05T07:07:40Z"}],"graph_snapshots":[{"event_id":"sha256:ba4761e8b5da1334f71ea1b2abed08f97c36328f0f52c0e52bf120d22fc73102","target":"graph","created_at":"2026-07-05T07:07:40Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2310.18512/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) often benefit from intermediate steps of reasoning to generate answers to complex problems. When these intermediate steps of reasoning are used to monitor the activity of the model, it is essential that this explicit reasoning is faithful, i.e. that it reflects what the model is actually reasoning about. In this work, we focus on one potential way intermediate steps of reasoning could be unfaithful: encoded reasoning, where an LLM could encode intermediate steps of reasoning in the generated text in a way that is not understandable to human readers. We show that la","authors_text":"Fabien Roger, Ryan Greenblatt","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T22:02:29Z","title":"Preventing Language Models From Hiding Their Reasoning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.18512","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:247a3aee97a5b2e9bcee3b8453992c834ee0cafd243c469a15416c42f070f7e8","target":"record","created_at":"2026-07-05T07:07:40Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"cb6a9831e895e3a02b78daa9ae61d223a1f19c6749bcd5dfd6be37ddc94dcb1f","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-27T22:02:29Z","title_canon_sha256":"59d0c8f3daf91c66baa3c2201d83d63d2a9f62e52b9eb1d23d07f90ea88f8810"},"schema_version":"1.0","source":{"id":"2310.18512","kind":"arxiv","version":2}},"canonical_sha256":"d9a2609b4c0c4156404f39fbf6e51a800819df044ef4b5d661fa6f17cfc79102","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d9a2609b4c0c4156404f39fbf6e51a800819df044ef4b5d661fa6f17cfc79102","first_computed_at":"2026-07-05T07:07:40.289697Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:07:40.289697Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"xmEij8+zXAIH51zdRGmaZKwc1yDqGPOyZm8Xum9/x27aAG9YMofpaTUW3H0JaMEh3Ot1OIqYyWzAWxfccbJHCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T07:07:40.290159Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.18512","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:247a3aee97a5b2e9bcee3b8453992c834ee0cafd243c469a15416c42f070f7e8","sha256:ba4761e8b5da1334f71ea1b2abed08f97c36328f0f52c0e52bf120d22fc73102"],"state_sha256":"a40250e6182d6f5525ac63d0a7927748dd2b4d8c2f9b69a0cdfcb7c715453fb3"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"qTQ8OUFOjhQiniDwvCPGYd/NRchDYf1MOFGRr5daqy1SABZjylh4zhSecSzVihoghvUS2YtORJNa+di0Zd6GAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T03:10:23.311429Z","bundle_sha256":"668d7889984d35d3c82272fec2ebd6137fbad313ef15d7fe43fee4e82bad093a"}}