{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GDD6QSU3F6MNBHEJQSEUYAOSMF","short_pith_number":"pith:GDD6QSU3","schema_version":"1.0","canonical_sha256":"30c7e84a9b2f98d09c8984894c01d2617fafac38a43f02816ce452adcd3c1865","source":{"kind":"arxiv","id":"2607.03640","version":1},"attestation_state":"computed","paper":{"title":"Revealing Hidden Model Behaviors with Task-Specific Self-Reports","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bartosz Zieli\\'nski, Taras Kutsyk","submitted_at":"2026-07-03T23:41:25Z","abstract_excerpt":"Fine-tuning can give a language model a hidden behavior--it may give false answers under a narrow condition, or give harmful advice only when a prompt touches a particular topic. We introduce the Stabilized Adapter for self-Report (SAR), a lightweight LoRA adapter that makes a fine-tuned model describe its own hidden behavior in plain language, using only the model and the dataset it was trained on. Across seven implanted behaviors (plus a no-behavior control), SAR detects the hidden behavior in every one--even when the model has generalized into broad misalignment that the training data alone"},"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":"2607.03640","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-07-03T23:41:25Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"b4a699e3fd96f04ba01ccd5e1bda715210cd153862e5f793326358c04266dec1","abstract_canon_sha256":"d4ea9febb9c2c0c59a909e8d4ef2cc5b1ce67b0da2f73d4ff1c738888774c920"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:17:58.954383Z","signature_b64":"72jJqELK5hMaDsjSJ6p5HmIbCsM708Il3RePvLxWpEzTmePvwGvnR1dzXv/ufcWrmjj843Eyq6w4CYmRYLoWBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"30c7e84a9b2f98d09c8984894c01d2617fafac38a43f02816ce452adcd3c1865","last_reissued_at":"2026-07-07T02:17:58.953660Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:17:58.953660Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Revealing Hidden Model Behaviors with Task-Specific Self-Reports","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bartosz Zieli\\'nski, Taras Kutsyk","submitted_at":"2026-07-03T23:41:25Z","abstract_excerpt":"Fine-tuning can give a language model a hidden behavior--it may give false answers under a narrow condition, or give harmful advice only when a prompt touches a particular topic. We introduce the Stabilized Adapter for self-Report (SAR), a lightweight LoRA adapter that makes a fine-tuned model describe its own hidden behavior in plain language, using only the model and the dataset it was trained on. Across seven implanted behaviors (plus a no-behavior control), SAR detects the hidden behavior in every one--even when the model has generalized into broad misalignment that the training data alone"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.03640","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/2607.03640/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":"2607.03640","created_at":"2026-07-07T02:17:58.953769+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.03640v1","created_at":"2026-07-07T02:17:58.953769+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.03640","created_at":"2026-07-07T02:17:58.953769+00:00"},{"alias_kind":"pith_short_12","alias_value":"GDD6QSU3F6MN","created_at":"2026-07-07T02:17:58.953769+00:00"},{"alias_kind":"pith_short_16","alias_value":"GDD6QSU3F6MNBHEJ","created_at":"2026-07-07T02:17:58.953769+00:00"},{"alias_kind":"pith_short_8","alias_value":"GDD6QSU3","created_at":"2026-07-07T02:17:58.953769+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/GDD6QSU3F6MNBHEJQSEUYAOSMF","json":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF.json","graph_json":"https://pith.science/api/pith-number/GDD6QSU3F6MNBHEJQSEUYAOSMF/graph.json","events_json":"https://pith.science/api/pith-number/GDD6QSU3F6MNBHEJQSEUYAOSMF/events.json","paper":"https://pith.science/paper/GDD6QSU3"},"agent_actions":{"view_html":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF","download_json":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF.json","view_paper":"https://pith.science/paper/GDD6QSU3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.03640&json=true","fetch_graph":"https://pith.science/api/pith-number/GDD6QSU3F6MNBHEJQSEUYAOSMF/graph.json","fetch_events":"https://pith.science/api/pith-number/GDD6QSU3F6MNBHEJQSEUYAOSMF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF/action/storage_attestation","attest_author":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF/action/author_attestation","sign_citation":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF/action/citation_signature","submit_replication":"https://pith.science/pith/GDD6QSU3F6MNBHEJQSEUYAOSMF/action/replication_record"}},"created_at":"2026-07-07T02:17:58.953769+00:00","updated_at":"2026-07-07T02:17:58.953769+00:00"}