{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JVHPLJH3NEVCMBNJ2ZR7QWKHVU","short_pith_number":"pith:JVHPLJH3","schema_version":"1.0","canonical_sha256":"4d4ef5a4fb692a2605a9d663f85947ad16e02daf45f4aa9faa49a91b138e739f","source":{"kind":"arxiv","id":"2606.11998","version":1},"attestation_state":"computed","paper":{"title":"Bootstrapped Monitoring: Leveraging Transparent Reasoning to Oversee Stronger AI Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Frank Xiao, Mary Phuong","submitted_at":"2026-06-10T12:24:29Z","abstract_excerpt":"Trusted monitoring is a cornerstone of AI control. However, as frontier models grow more capable, the increasing capabilities gap between trusted and untrusted models may render trusted models unreliable monitors. We introduce \\emph{bootstrapped monitoring}, a protocol that addresses this by inserting a stronger, intermediate untrusted model with transparent chain-of-thought reasoning into the oversight chain. The untrusted monitor ($U_m$) evaluates the agent's actions, while a weaker trusted model ($T$) oversees $U_m$'s reasoning to detect collusion. We evaluate bootstrapped monitoring on mul"},"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":"2606.11998","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-10T12:24:29Z","cross_cats_sorted":[],"title_canon_sha256":"99727cd9c94f936ec9128d4a64cdbe7088702e54b75b3fe79bd63dfdecfcdcbf","abstract_canon_sha256":"8a378fcb12e1f4fad119e391b10c1c5ad338649bfa52172e247e23a16bcefb14"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-11T01:10:41.619000Z","signature_b64":"k66RZuZADKn1/LQt+eIIPHtFXxCDUcKioD2N36Qwnzm3BmRTdhY8MBVWE7BhHzAYh6gY7uRCOK+XaE9MyhXxDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4d4ef5a4fb692a2605a9d663f85947ad16e02daf45f4aa9faa49a91b138e739f","last_reissued_at":"2026-06-11T01:10:41.618197Z","signature_status":"signed_v1","first_computed_at":"2026-06-11T01:10:41.618197Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bootstrapped Monitoring: Leveraging Transparent Reasoning to Oversee Stronger AI Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Frank Xiao, Mary Phuong","submitted_at":"2026-06-10T12:24:29Z","abstract_excerpt":"Trusted monitoring is a cornerstone of AI control. However, as frontier models grow more capable, the increasing capabilities gap between trusted and untrusted models may render trusted models unreliable monitors. We introduce \\emph{bootstrapped monitoring}, a protocol that addresses this by inserting a stronger, intermediate untrusted model with transparent chain-of-thought reasoning into the oversight chain. The untrusted monitor ($U_m$) evaluates the agent's actions, while a weaker trusted model ($T$) oversees $U_m$'s reasoning to detect collusion. We evaluate bootstrapped monitoring on mul"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.11998","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/2606.11998/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":"2606.11998","created_at":"2026-06-11T01:10:41.618323+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.11998v1","created_at":"2026-06-11T01:10:41.618323+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.11998","created_at":"2026-06-11T01:10:41.618323+00:00"},{"alias_kind":"pith_short_12","alias_value":"JVHPLJH3NEVC","created_at":"2026-06-11T01:10:41.618323+00:00"},{"alias_kind":"pith_short_16","alias_value":"JVHPLJH3NEVCMBNJ","created_at":"2026-06-11T01:10:41.618323+00:00"},{"alias_kind":"pith_short_8","alias_value":"JVHPLJH3","created_at":"2026-06-11T01:10:41.618323+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/JVHPLJH3NEVCMBNJ2ZR7QWKHVU","json":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU.json","graph_json":"https://pith.science/api/pith-number/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/graph.json","events_json":"https://pith.science/api/pith-number/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/events.json","paper":"https://pith.science/paper/JVHPLJH3"},"agent_actions":{"view_html":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU","download_json":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU.json","view_paper":"https://pith.science/paper/JVHPLJH3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.11998&json=true","fetch_graph":"https://pith.science/api/pith-number/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/graph.json","fetch_events":"https://pith.science/api/pith-number/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/action/storage_attestation","attest_author":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/action/author_attestation","sign_citation":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/action/citation_signature","submit_replication":"https://pith.science/pith/JVHPLJH3NEVCMBNJ2ZR7QWKHVU/action/replication_record"}},"created_at":"2026-06-11T01:10:41.618323+00:00","updated_at":"2026-06-11T01:10:41.618323+00:00"}