{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:BXYPV5FBFSSSBDA6J36H5XKVFZ","short_pith_number":"pith:BXYPV5FB","canonical_record":{"source":{"id":"2504.17130","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-23T22:47:30Z","cross_cats_sorted":["cs.CR","cs.CY"],"title_canon_sha256":"0a37f3f102fabed14b65a5282a755f44d02defd0841a3e58a4311942b4778c71","abstract_canon_sha256":"d49f3f0aa563d3a325f4b6a583a9f6431412f9adecbc64e2d2541cbfaae8f3eb"},"schema_version":"1.0"},"canonical_sha256":"0df0faf4a12ca5208c1e4efc7edd552e713cb469729994a05ebca82cd24bca87","source":{"kind":"arxiv","id":"2504.17130","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.17130","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"arxiv_version","alias_value":"2504.17130v3","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.17130","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"pith_short_12","alias_value":"BXYPV5FBFSSS","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"pith_short_16","alias_value":"BXYPV5FBFSSSBDA6","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"pith_short_8","alias_value":"BXYPV5FB","created_at":"2026-07-05T11:52:02Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:BXYPV5FBFSSSBDA6J36H5XKVFZ","target":"record","payload":{"canonical_record":{"source":{"id":"2504.17130","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-23T22:47:30Z","cross_cats_sorted":["cs.CR","cs.CY"],"title_canon_sha256":"0a37f3f102fabed14b65a5282a755f44d02defd0841a3e58a4311942b4778c71","abstract_canon_sha256":"d49f3f0aa563d3a325f4b6a583a9f6431412f9adecbc64e2d2541cbfaae8f3eb"},"schema_version":"1.0"},"canonical_sha256":"0df0faf4a12ca5208c1e4efc7edd552e713cb469729994a05ebca82cd24bca87","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:52:02.053770Z","signature_b64":"TtNO26nV2d3rGKE3/VuImGBPW+FjXpiFrFtjPcCFO/1tzViCGJMNORu7xhV9Xxw91fWGytIeCn4Q503zXXlzDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0df0faf4a12ca5208c1e4efc7edd552e713cb469729994a05ebca82cd24bca87","last_reissued_at":"2026-07-05T11:52:02.053309Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:52:02.053309Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2504.17130","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:52:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AxHVAEZXcA4uGxQMcQyBxGOrvF1UmLn85WE167eObZeO9v5riIqGYNtFcPuJeB05i5jcFmthjkeN5PDjjnV5AQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T05:53:18.214848Z"},"content_sha256":"bc193c2666861afccf47d2ba320d0931a29cb12612f1e1ca344aef927f739bda","schema_version":"1.0","event_id":"sha256:bc193c2666861afccf47d2ba320d0931a29cb12612f1e1ca344aef927f739bda"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:BXYPV5FBFSSSBDA6J36H5XKVFZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Steering the CensorShip: Uncovering Representation Vectors for LLM \"Thought\" Control","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR","cs.CY"],"primary_cat":"cs.CL","authors_text":"David Evans, Hannah Cyberey","submitted_at":"2025-04-23T22:47:30Z","abstract_excerpt":"Large language models (LLMs) have transformed the way we access information. These models are often tuned to refuse to comply with requests that are considered harmful and to produce responses that better align with the preferences of those who control the models. To understand how this \"censorship\" works. We use representation engineering techniques to study open-weights safety-tuned models. We present a method for finding a refusal--compliance vector that detects and controls the level of censorship in model outputs. We also analyze recent reasoning LLMs, distilled from DeepSeek-R1, and unco"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.17130","kind":"arxiv","version":3},"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/2504.17130/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":"2026-07-05T11:52:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/pt8zR1nwEpqB/BnF+2ratMkO/jN1nLVaMDzmRBf8M+1MmVaBr1b8Yt3F2kBlp/HAsot4s0FuAYB3M8qgbv2DQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T05:53:18.215364Z"},"content_sha256":"02c570ede6cfe70cc343898b2f2ec5b49245e3a49b18012a31a6d86f1c0b1390","schema_version":"1.0","event_id":"sha256:02c570ede6cfe70cc343898b2f2ec5b49245e3a49b18012a31a6d86f1c0b1390"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ/bundle.json","state_url":"https://pith.science/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ/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":"2026-08-10T05:53:18Z","links":{"resolver":"https://pith.science/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ","bundle":"https://pith.science/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ/bundle.json","state":"https://pith.science/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/BXYPV5FBFSSSBDA6J36H5XKVFZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:BXYPV5FBFSSSBDA6J36H5XKVFZ","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":"d49f3f0aa563d3a325f4b6a583a9f6431412f9adecbc64e2d2541cbfaae8f3eb","cross_cats_sorted":["cs.CR","cs.CY"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-23T22:47:30Z","title_canon_sha256":"0a37f3f102fabed14b65a5282a755f44d02defd0841a3e58a4311942b4778c71"},"schema_version":"1.0","source":{"id":"2504.17130","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.17130","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"arxiv_version","alias_value":"2504.17130v3","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.17130","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"pith_short_12","alias_value":"BXYPV5FBFSSS","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"pith_short_16","alias_value":"BXYPV5FBFSSSBDA6","created_at":"2026-07-05T11:52:02Z"},{"alias_kind":"pith_short_8","alias_value":"BXYPV5FB","created_at":"2026-07-05T11:52:02Z"}],"graph_snapshots":[{"event_id":"sha256:02c570ede6cfe70cc343898b2f2ec5b49245e3a49b18012a31a6d86f1c0b1390","target":"graph","created_at":"2026-07-05T11:52:02Z","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/2504.17130/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) have transformed the way we access information. These models are often tuned to refuse to comply with requests that are considered harmful and to produce responses that better align with the preferences of those who control the models. To understand how this \"censorship\" works. We use representation engineering techniques to study open-weights safety-tuned models. We present a method for finding a refusal--compliance vector that detects and controls the level of censorship in model outputs. We also analyze recent reasoning LLMs, distilled from DeepSeek-R1, and unco","authors_text":"David Evans, Hannah Cyberey","cross_cats":["cs.CR","cs.CY"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-23T22:47:30Z","title":"Steering the CensorShip: Uncovering Representation Vectors for LLM \"Thought\" Control"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.17130","kind":"arxiv","version":3},"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:bc193c2666861afccf47d2ba320d0931a29cb12612f1e1ca344aef927f739bda","target":"record","created_at":"2026-07-05T11:52:02Z","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":"d49f3f0aa563d3a325f4b6a583a9f6431412f9adecbc64e2d2541cbfaae8f3eb","cross_cats_sorted":["cs.CR","cs.CY"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-23T22:47:30Z","title_canon_sha256":"0a37f3f102fabed14b65a5282a755f44d02defd0841a3e58a4311942b4778c71"},"schema_version":"1.0","source":{"id":"2504.17130","kind":"arxiv","version":3}},"canonical_sha256":"0df0faf4a12ca5208c1e4efc7edd552e713cb469729994a05ebca82cd24bca87","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0df0faf4a12ca5208c1e4efc7edd552e713cb469729994a05ebca82cd24bca87","first_computed_at":"2026-07-05T11:52:02.053309Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:52:02.053309Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"TtNO26nV2d3rGKE3/VuImGBPW+FjXpiFrFtjPcCFO/1tzViCGJMNORu7xhV9Xxw91fWGytIeCn4Q503zXXlzDw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:52:02.053770Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.17130","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bc193c2666861afccf47d2ba320d0931a29cb12612f1e1ca344aef927f739bda","sha256:02c570ede6cfe70cc343898b2f2ec5b49245e3a49b18012a31a6d86f1c0b1390"],"state_sha256":"6da787c95b89320f02ec1e472c0cbcdf6211ec641d100e442f1d7842aad9d21e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LPAOutwNexRQlS/qK9SHd9dVN664C6zw8ZmRuXZSmNfjGhBYEpp/vDB5tN0fMcrezTYhwatOdDiPusi+3Tm8DA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-10T05:53:18.219260Z","bundle_sha256":"de7c75d8c0b84139be1f002d2771f5168da07f2ad5fc76bd20d93a90ca1e3167"}}