{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:VTQABSCJ5TUZH544TI5ZWHRRT6","short_pith_number":"pith:VTQABSCJ","canonical_record":{"source":{"id":"2405.19550","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-29T22:26:26Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"0928cee1eb8b8d1fd05affa88ba487b9ff423730818061ac02ba6ac98f405ff4","abstract_canon_sha256":"d4c2c5844f7ac4f5fcb00a1bd0c16ee74b402660674c41f8ae618f0b18e2b17f"},"schema_version":"1.0"},"canonical_sha256":"ace000c849ece993f79c9a3b9b1e319f9a1de0148216df3154a5d94640a02d58","source":{"kind":"arxiv","id":"2405.19550","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.19550","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"arxiv_version","alias_value":"2405.19550v1","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.19550","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"pith_short_12","alias_value":"VTQABSCJ5TUZ","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"pith_short_16","alias_value":"VTQABSCJ5TUZH544","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"pith_short_8","alias_value":"VTQABSCJ","created_at":"2026-07-05T08:25:23Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:VTQABSCJ5TUZH544TI5ZWHRRT6","target":"record","payload":{"canonical_record":{"source":{"id":"2405.19550","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-29T22:26:26Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"0928cee1eb8b8d1fd05affa88ba487b9ff423730818061ac02ba6ac98f405ff4","abstract_canon_sha256":"d4c2c5844f7ac4f5fcb00a1bd0c16ee74b402660674c41f8ae618f0b18e2b17f"},"schema_version":"1.0"},"canonical_sha256":"ace000c849ece993f79c9a3b9b1e319f9a1de0148216df3154a5d94640a02d58","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:25:23.407819Z","signature_b64":"IcPIxVXIMIPkikLGn2FUCsSYH3LMtmW+EdSzBh/p2iRr3wMFUFvo1a8UyD3U+aYbjORxLQ/WVJVQQyH4a6CdAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ace000c849ece993f79c9a3b9b1e319f9a1de0148216df3154a5d94640a02d58","last_reissued_at":"2026-07-05T08:25:23.407386Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:25:23.407386Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.19550","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:25:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vjQkZaV0iCyxBjcr5gzOZpxpeS4RYXnFDXZ82q0JIRU2iGI0aJnCHp8YuYaBrAsm/FArt+bo00zGtrHhZ8q2Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T08:34:37.098054Z"},"content_sha256":"02ab5949014a9767f9536a671fa1f45cb49cbcb200d6ede061c23b95b400b04b","schema_version":"1.0","event_id":"sha256:02ab5949014a9767f9536a671fa1f45cb49cbcb200d6ede061c23b95b400b04b"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:VTQABSCJ5TUZH544TI5ZWHRRT6","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Stress-Testing Capability Elicitation With Password-Locked Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"David Krueger, Dmitrii Krasheninnikov, Fabien Roger, Ryan Greenblatt","submitted_at":"2024-05-29T22:26:26Z","abstract_excerpt":"To determine the safety of large language models (LLMs), AI developers must be able to assess their dangerous capabilities. But simple prompting strategies often fail to elicit an LLM's full capabilities. One way to elicit capabilities more robustly is to fine-tune the LLM to complete the task. In this paper, we investigate the conditions under which fine-tuning-based elicitation suffices to elicit capabilities. To do this, we introduce password-locked models, LLMs fine-tuned such that some of their capabilities are deliberately hidden. Specifically, these LLMs are trained to exhibit these cap"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.19550","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/2405.19550/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":"2026-07-05T08:25:23Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QK85dy0Tfep2XGrZDtPW5rVl4UhjyJGWIWN0RndbmF0Ju1GtT8n5YWeM3w/X4W6j3Iv28Hd0Ubmo/OiTFadLBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T08:34:37.098894Z"},"content_sha256":"33fe42f206259dcd23283986c4abe8dfe825b8743f32ea371c3a1b2e911826c4","schema_version":"1.0","event_id":"sha256:33fe42f206259dcd23283986c4abe8dfe825b8743f32ea371c3a1b2e911826c4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/VTQABSCJ5TUZH544TI5ZWHRRT6/bundle.json","state_url":"https://pith.science/pith/VTQABSCJ5TUZH544TI5ZWHRRT6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/VTQABSCJ5TUZH544TI5ZWHRRT6/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":"2026-08-06T08:34:37Z","links":{"resolver":"https://pith.science/pith/VTQABSCJ5TUZH544TI5ZWHRRT6","bundle":"https://pith.science/pith/VTQABSCJ5TUZH544TI5ZWHRRT6/bundle.json","state":"https://pith.science/pith/VTQABSCJ5TUZH544TI5ZWHRRT6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/VTQABSCJ5TUZH544TI5ZWHRRT6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:VTQABSCJ5TUZH544TI5ZWHRRT6","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":"d4c2c5844f7ac4f5fcb00a1bd0c16ee74b402660674c41f8ae618f0b18e2b17f","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-29T22:26:26Z","title_canon_sha256":"0928cee1eb8b8d1fd05affa88ba487b9ff423730818061ac02ba6ac98f405ff4"},"schema_version":"1.0","source":{"id":"2405.19550","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.19550","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"arxiv_version","alias_value":"2405.19550v1","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.19550","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"pith_short_12","alias_value":"VTQABSCJ5TUZ","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"pith_short_16","alias_value":"VTQABSCJ5TUZH544","created_at":"2026-07-05T08:25:23Z"},{"alias_kind":"pith_short_8","alias_value":"VTQABSCJ","created_at":"2026-07-05T08:25:23Z"}],"graph_snapshots":[{"event_id":"sha256:33fe42f206259dcd23283986c4abe8dfe825b8743f32ea371c3a1b2e911826c4","target":"graph","created_at":"2026-07-05T08:25:23Z","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/2405.19550/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"To determine the safety of large language models (LLMs), AI developers must be able to assess their dangerous capabilities. But simple prompting strategies often fail to elicit an LLM's full capabilities. One way to elicit capabilities more robustly is to fine-tune the LLM to complete the task. In this paper, we investigate the conditions under which fine-tuning-based elicitation suffices to elicit capabilities. To do this, we introduce password-locked models, LLMs fine-tuned such that some of their capabilities are deliberately hidden. Specifically, these LLMs are trained to exhibit these cap","authors_text":"David Krueger, Dmitrii Krasheninnikov, Fabien Roger, Ryan Greenblatt","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-29T22:26:26Z","title":"Stress-Testing Capability Elicitation With Password-Locked Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.19550","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":"sha256:02ab5949014a9767f9536a671fa1f45cb49cbcb200d6ede061c23b95b400b04b","target":"record","created_at":"2026-07-05T08:25:23Z","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":"d4c2c5844f7ac4f5fcb00a1bd0c16ee74b402660674c41f8ae618f0b18e2b17f","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-29T22:26:26Z","title_canon_sha256":"0928cee1eb8b8d1fd05affa88ba487b9ff423730818061ac02ba6ac98f405ff4"},"schema_version":"1.0","source":{"id":"2405.19550","kind":"arxiv","version":1}},"canonical_sha256":"ace000c849ece993f79c9a3b9b1e319f9a1de0148216df3154a5d94640a02d58","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ace000c849ece993f79c9a3b9b1e319f9a1de0148216df3154a5d94640a02d58","first_computed_at":"2026-07-05T08:25:23.407386Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:25:23.407386Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"IcPIxVXIMIPkikLGn2FUCsSYH3LMtmW+EdSzBh/p2iRr3wMFUFvo1a8UyD3U+aYbjORxLQ/WVJVQQyH4a6CdAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:25:23.407819Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.19550","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:02ab5949014a9767f9536a671fa1f45cb49cbcb200d6ede061c23b95b400b04b","sha256:33fe42f206259dcd23283986c4abe8dfe825b8743f32ea371c3a1b2e911826c4"],"state_sha256":"c25a1386942cf376988f2ac1c30867ab0ea1a0ce695587221c3cd3a9354f5275"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"c6ldtlt9GSeN7IufRluYqRHcz421Cv0VcoKE+yiBGZcrYRO0RR4jAkkl4QAFnqIP7zPlTIgc/Gzn/k/HC2LaDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T08:34:37.103204Z","bundle_sha256":"1a0bde327efe175bc17d22f0d1d7cebaf14484ae2a88cf32526ca011db3ad4e6"}}