{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7JXN63X6FP6M3KRIMXAWL3OZYY","short_pith_number":"pith:7JXN63X6","schema_version":"1.0","canonical_sha256":"fa6edf6efe2bfccdaa2865c165edd9c616c3e10823830f18d488dc297d061bf4","source":{"kind":"arxiv","id":"2608.04735","version":1},"attestation_state":"computed","paper":{"title":"Chain-of-Thought Monitoring Can Be Unreliable in Implicit-Influence Settings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Agatha Duzan, Asa Cooper Stickland","submitted_at":"2026-08-05T11:59:20Z","abstract_excerpt":"Chain-of-thought (CoT) monitoring is increasingly treated as an important safety layer for frontier reasoning models. Most monitorability evaluations study explicit-influence settings: setups where the prompt directly incentivizes the model to hide something, e.g., by instructing it to perform a hidden side-task. A complementary axis for CoT-monitor evaluations is implicit-influence settings, where the prompt contains no instruction to hide, but the model's behavior is still shaped by features of the task or context, e.g. an irrelevant detail about a candidate that biases a hiring rating.\n  We"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2608.04735","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-05T11:59:20Z","cross_cats_sorted":[],"title_canon_sha256":"780d2425b1dfdc469542e0e65d2bd7b35b26ea219ff43446168244cd5abb404a","abstract_canon_sha256":"23a0d9d60616d9e61b9cabf82afeb2113590687adca05d52522e24f0a7d2a7b4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-06T01:47:00.049008Z","signature_b64":"YNm1LdbRwPEkdJ3wjdpOikqkL//IPNyH6sGj973nIUNUP84n5E1vjqevz2ELAVl7R7sMbk5X7/6UsMggfD5MCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fa6edf6efe2bfccdaa2865c165edd9c616c3e10823830f18d488dc297d061bf4","last_reissued_at":"2026-08-06T01:47:00.047499Z","signature_status":"signed_v1","first_computed_at":"2026-08-06T01:47:00.047499Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Chain-of-Thought Monitoring Can Be Unreliable in Implicit-Influence Settings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Agatha Duzan, Asa Cooper Stickland","submitted_at":"2026-08-05T11:59:20Z","abstract_excerpt":"Chain-of-thought (CoT) monitoring is increasingly treated as an important safety layer for frontier reasoning models. Most monitorability evaluations study explicit-influence settings: setups where the prompt directly incentivizes the model to hide something, e.g., by instructing it to perform a hidden side-task. A complementary axis for CoT-monitor evaluations is implicit-influence settings, where the prompt contains no instruction to hide, but the model's behavior is still shaped by features of the task or context, e.g. an irrelevant detail about a candidate that biases a hiring rating.\n  We"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.04735","kind":"arxiv","version":1},"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/2608.04735/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2608.04735","created_at":"2026-08-06T01:47:00.049382+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.04735v1","created_at":"2026-08-06T01:47:00.049382+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.04735","created_at":"2026-08-06T01:47:00.049382+00:00"},{"alias_kind":"pith_short_12","alias_value":"7JXN63X6FP6M","created_at":"2026-08-06T01:47:00.049382+00:00"},{"alias_kind":"pith_short_16","alias_value":"7JXN63X6FP6M3KRI","created_at":"2026-08-06T01:47:00.049382+00:00"},{"alias_kind":"pith_short_8","alias_value":"7JXN63X6","created_at":"2026-08-06T01:47:00.049382+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY","json":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY.json","graph_json":"https://pith.science/api/pith-number/7JXN63X6FP6M3KRIMXAWL3OZYY/graph.json","events_json":"https://pith.science/api/pith-number/7JXN63X6FP6M3KRIMXAWL3OZYY/events.json","paper":"https://pith.science/paper/7JXN63X6"},"agent_actions":{"view_html":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY","download_json":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY.json","view_paper":"https://pith.science/paper/7JXN63X6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.04735&json=true","fetch_graph":"https://pith.science/api/pith-number/7JXN63X6FP6M3KRIMXAWL3OZYY/graph.json","fetch_events":"https://pith.science/api/pith-number/7JXN63X6FP6M3KRIMXAWL3OZYY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY/action/storage_attestation","attest_author":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY/action/author_attestation","sign_citation":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY/action/citation_signature","submit_replication":"https://pith.science/pith/7JXN63X6FP6M3KRIMXAWL3OZYY/action/replication_record"}},"created_at":"2026-08-06T01:47:00.049382+00:00","updated_at":"2026-08-06T01:47:00.049382+00:00"}