{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:Q5HZRNCCBR23VZVFCSQNJLVJIR","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":"5acab73228edf748b92442cbd2e84cbac7ee0044354bc7be16ad552fc4d4e4c2","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-05-24T17:41:47Z","title_canon_sha256":"1e25620bee427f14aff080f8cbf4b9cdc9edd0a6296589341cf05ea179c0fbb4"},"schema_version":"1.0","source":{"id":"2505.18807","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.18807","created_at":"2026-07-05T11:09:19Z"},{"alias_kind":"arxiv_version","alias_value":"2505.18807v1","created_at":"2026-07-05T11:09:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.18807","created_at":"2026-07-05T11:09:19Z"},{"alias_kind":"pith_short_12","alias_value":"Q5HZRNCCBR23","created_at":"2026-07-05T11:09:19Z"},{"alias_kind":"pith_short_16","alias_value":"Q5HZRNCCBR23VZVF","created_at":"2026-07-05T11:09:19Z"},{"alias_kind":"pith_short_8","alias_value":"Q5HZRNCC","created_at":"2026-07-05T11:09:19Z"}],"graph_snapshots":[{"event_id":"sha256:d7bd88bf57a5bbc35c30ac4f6a876df1f8024785045a4973ec4940a6c9b3c159","target":"graph","created_at":"2026-07-05T11:09:19Z","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/2505.18807/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Modern large language models rely on chain-of-thought (CoT) reasoning to achieve impressive performance, yet the same mechanism can amplify deceptive alignment, situations in which a model appears aligned while covertly pursuing misaligned goals. Existing safety pipelines treat deception as a black-box output to be filtered post-hoc, leaving the model free to scheme during its internal reasoning. We ask: Can deception be intercepted while the model is thinking? We answer this question, the first framework that embeds a Self-Monitor inside the CoT process itself, named CoT Monitor+. During gene","authors_text":"Boyuan Chen, Donghai Hong, Jiaming Ji, Jiayi Zhou, Juntao Dai, Kaile Wang, Sirui Han, Sitong Fang, Wenqi Chen, Yaodong Yang, Yike Guo","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-05-24T17:41:47Z","title":"Mitigating Deceptive Alignment via Self-Monitoring"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.18807","kind":"arxiv","version":1},"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":"sha256:70937a7a9097628a6efaa16913b3728cca63d76c3a594f46e33d83d6cb998153","target":"record","created_at":"2026-07-05T11:09:19Z","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":"5acab73228edf748b92442cbd2e84cbac7ee0044354bc7be16ad552fc4d4e4c2","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-05-24T17:41:47Z","title_canon_sha256":"1e25620bee427f14aff080f8cbf4b9cdc9edd0a6296589341cf05ea179c0fbb4"},"schema_version":"1.0","source":{"id":"2505.18807","kind":"arxiv","version":1}},"canonical_sha256":"874f98b4420c75bae6a514a0d4aea94453712bbad5e6ffa4741dfbdf1caea8e9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"874f98b4420c75bae6a514a0d4aea94453712bbad5e6ffa4741dfbdf1caea8e9","first_computed_at":"2026-07-05T11:09:19.951052Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:09:19.951052Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"+6IAT+Y3Zrq4etiKEJSBIV5+zdpsZywX6FEBrXS1LekMGiM9U5ymzGW+1djIf4egZruSutPE0ch0iow+kJlSDg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:09:19.951599Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.18807","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:70937a7a9097628a6efaa16913b3728cca63d76c3a594f46e33d83d6cb998153","sha256:d7bd88bf57a5bbc35c30ac4f6a876df1f8024785045a4973ec4940a6c9b3c159"],"state_sha256":"73b26bea44333f81b3515b1989997ccf46bec160b3544c537ba4039085a543f1"}