{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:IXH56MIJEPLGDXQUVQNTN36MZY","short_pith_number":"pith:IXH56MIJ","canonical_record":{"source":{"id":"2312.06942","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-12T02:34:06Z","cross_cats_sorted":[],"title_canon_sha256":"a52a2846fb875879ad6a71f0acf4df511643a9a9f06b58dde6ba44a5c96c93a3","abstract_canon_sha256":"dc69141ab974b8dae420924735e9c63de179e7f5811c50f85d142f9a9a9d1655"},"schema_version":"1.0"},"canonical_sha256":"45cfdf310923d661de14ac1b36efccce3674b84c7adc7501cf241e2d32268356","source":{"kind":"arxiv","id":"2312.06942","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.06942","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"arxiv_version","alias_value":"2312.06942v5","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.06942","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"pith_short_12","alias_value":"IXH56MIJEPLG","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"pith_short_16","alias_value":"IXH56MIJEPLGDXQU","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"pith_short_8","alias_value":"IXH56MIJ","created_at":"2026-07-05T08:47:09Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:IXH56MIJEPLGDXQUVQNTN36MZY","target":"record","payload":{"canonical_record":{"source":{"id":"2312.06942","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-12T02:34:06Z","cross_cats_sorted":[],"title_canon_sha256":"a52a2846fb875879ad6a71f0acf4df511643a9a9f06b58dde6ba44a5c96c93a3","abstract_canon_sha256":"dc69141ab974b8dae420924735e9c63de179e7f5811c50f85d142f9a9a9d1655"},"schema_version":"1.0"},"canonical_sha256":"45cfdf310923d661de14ac1b36efccce3674b84c7adc7501cf241e2d32268356","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:47:09.252374Z","signature_b64":"zYCtR+h7FFi6leKIf+VyiZ/B3wfIy6YJYXp3BAPi40g1yaA2PQKej+Xb3apBjxCrPqNxkdxsgGxsovp6z/m4BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45cfdf310923d661de14ac1b36efccce3674b84c7adc7501cf241e2d32268356","last_reissued_at":"2026-07-05T08:47:09.251835Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:47:09.251835Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2312.06942","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:47:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ayUR1xXUEjLdBGAuAVUI0TKteCTG5FruDYHLZ1wAqjoRUqB/k626f85v0xQ7pYa5ErPcpaYO/bUwjb4MdMmjDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-18T19:08:17.724461Z"},"content_sha256":"19e8ff4e9c1739e79770315c360a61435c79c53c3963dcb5a65bbf973dac8950","schema_version":"1.0","event_id":"sha256:19e8ff4e9c1739e79770315c360a61435c79c53c3963dcb5a65bbf973dac8950"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:IXH56MIJEPLGDXQUVQNTN36MZY","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"AI Control: Improving Safety Despite Intentional Subversion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Buck Shlegeris, Fabien Roger, Kshitij Sachan, Ryan Greenblatt","submitted_at":"2023-12-12T02:34:06Z","abstract_excerpt":"As large language models (LLMs) become more powerful and are deployed more autonomously, it will be increasingly important to prevent them from causing harmful outcomes. Researchers have investigated a variety of safety techniques for this purpose, e.g. using models to review the outputs of other models, or red-teaming techniques to surface subtle failure modes. However, researchers have not evaluated whether such techniques still ensure safety if the model is itself intentionally trying to subvert them. In this paper, we develop and evaluate pipelines of safety techniques (\"protocols\") that a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.06942","kind":"arxiv","version":5},"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/2312.06942/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":"2026-07-05T08:47:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"EWe1xYMM0K0q2ffn8fIFWgZ5kqIAiJoqM7vAnrdYH8e/WLsnX/QMkVbf0PWcdEAhm5UHBZs/37kUDkXG9JDdAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-18T19:08:17.724842Z"},"content_sha256":"f64dcbc786f06f2aa320b5cc2ae906350c0d6b3d8b4128a20f0635c95206147d","schema_version":"1.0","event_id":"sha256:f64dcbc786f06f2aa320b5cc2ae906350c0d6b3d8b4128a20f0635c95206147d"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IXH56MIJEPLGDXQUVQNTN36MZY/bundle.json","state_url":"https://pith.science/pith/IXH56MIJEPLGDXQUVQNTN36MZY/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IXH56MIJEPLGDXQUVQNTN36MZY/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":"2026-07-18T19:08:17Z","links":{"resolver":"https://pith.science/pith/IXH56MIJEPLGDXQUVQNTN36MZY","bundle":"https://pith.science/pith/IXH56MIJEPLGDXQUVQNTN36MZY/bundle.json","state":"https://pith.science/pith/IXH56MIJEPLGDXQUVQNTN36MZY/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IXH56MIJEPLGDXQUVQNTN36MZY/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:IXH56MIJEPLGDXQUVQNTN36MZY","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":"dc69141ab974b8dae420924735e9c63de179e7f5811c50f85d142f9a9a9d1655","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-12T02:34:06Z","title_canon_sha256":"a52a2846fb875879ad6a71f0acf4df511643a9a9f06b58dde6ba44a5c96c93a3"},"schema_version":"1.0","source":{"id":"2312.06942","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.06942","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"arxiv_version","alias_value":"2312.06942v5","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.06942","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"pith_short_12","alias_value":"IXH56MIJEPLG","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"pith_short_16","alias_value":"IXH56MIJEPLGDXQU","created_at":"2026-07-05T08:47:09Z"},{"alias_kind":"pith_short_8","alias_value":"IXH56MIJ","created_at":"2026-07-05T08:47:09Z"}],"graph_snapshots":[{"event_id":"sha256:f64dcbc786f06f2aa320b5cc2ae906350c0d6b3d8b4128a20f0635c95206147d","target":"graph","created_at":"2026-07-05T08:47:09Z","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/2312.06942/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"As large language models (LLMs) become more powerful and are deployed more autonomously, it will be increasingly important to prevent them from causing harmful outcomes. Researchers have investigated a variety of safety techniques for this purpose, e.g. using models to review the outputs of other models, or red-teaming techniques to surface subtle failure modes. However, researchers have not evaluated whether such techniques still ensure safety if the model is itself intentionally trying to subvert them. In this paper, we develop and evaluate pipelines of safety techniques (\"protocols\") that a","authors_text":"Buck Shlegeris, Fabien Roger, Kshitij Sachan, Ryan Greenblatt","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-12T02:34:06Z","title":"AI Control: Improving Safety Despite Intentional Subversion"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.06942","kind":"arxiv","version":5},"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":"sha256:19e8ff4e9c1739e79770315c360a61435c79c53c3963dcb5a65bbf973dac8950","target":"record","created_at":"2026-07-05T08:47:09Z","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":"dc69141ab974b8dae420924735e9c63de179e7f5811c50f85d142f9a9a9d1655","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-12T02:34:06Z","title_canon_sha256":"a52a2846fb875879ad6a71f0acf4df511643a9a9f06b58dde6ba44a5c96c93a3"},"schema_version":"1.0","source":{"id":"2312.06942","kind":"arxiv","version":5}},"canonical_sha256":"45cfdf310923d661de14ac1b36efccce3674b84c7adc7501cf241e2d32268356","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"45cfdf310923d661de14ac1b36efccce3674b84c7adc7501cf241e2d32268356","first_computed_at":"2026-07-05T08:47:09.251835Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:47:09.251835Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"zYCtR+h7FFi6leKIf+VyiZ/B3wfIy6YJYXp3BAPi40g1yaA2PQKej+Xb3apBjxCrPqNxkdxsgGxsovp6z/m4BQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:47:09.252374Z","signed_message":"canonical_sha256_bytes"},"source_id":"2312.06942","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:19e8ff4e9c1739e79770315c360a61435c79c53c3963dcb5a65bbf973dac8950","sha256:f64dcbc786f06f2aa320b5cc2ae906350c0d6b3d8b4128a20f0635c95206147d"],"state_sha256":"4acb0542096379dbeda553e91771307b271929f6103cd3f2564c362d451a735d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7MwbbDSsZBuwPfp+K0fyCSYtJs0NidwWW6/YZpe3+FwSkrYxWfOiFo6BONF/1ybVbRmB5HbE8qQJCyS/mLMwBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-18T19:08:17.726961Z","bundle_sha256":"67a592d02348dcd293f9054c8a9ea5fce968480112e77c9d0cc5cd6bd00432cb"}}