{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:ARMCGDQE2V5KDGL7FQQZJCWMYG","short_pith_number":"pith:ARMCGDQE","canonical_record":{"source":{"id":"2510.26418","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-10-30T12:10:03Z","cross_cats_sorted":[],"title_canon_sha256":"52e93352d29bc904afed2c494aa8028596e71729c2e9167cf9fef3e50293d3b9","abstract_canon_sha256":"5795eb1cda8c62ec4b3e15395dc3f2cccfeb4635e5153e1c1d5f25cc44c6d041"},"schema_version":"1.0"},"canonical_sha256":"0458230e04d57aa1997f2c21948accc1965dbeb706e9499129f15696d1269661","source":{"kind":"arxiv","id":"2510.26418","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2510.26418","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"arxiv_version","alias_value":"2510.26418v4","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.26418","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"pith_short_12","alias_value":"ARMCGDQE2V5K","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"pith_short_16","alias_value":"ARMCGDQE2V5KDGL7","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"pith_short_8","alias_value":"ARMCGDQE","created_at":"2026-05-26T01:03:18Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:ARMCGDQE2V5KDGL7FQQZJCWMYG","target":"record","payload":{"canonical_record":{"source":{"id":"2510.26418","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-10-30T12:10:03Z","cross_cats_sorted":[],"title_canon_sha256":"52e93352d29bc904afed2c494aa8028596e71729c2e9167cf9fef3e50293d3b9","abstract_canon_sha256":"5795eb1cda8c62ec4b3e15395dc3f2cccfeb4635e5153e1c1d5f25cc44c6d041"},"schema_version":"1.0"},"canonical_sha256":"0458230e04d57aa1997f2c21948accc1965dbeb706e9499129f15696d1269661","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:03:18.314117Z","signature_b64":"NQid23hMHAnrBqQfHR+ILgNrkaYLV+OjSj5UVkEuMBkFQoq+VfVNj28X0/iEwl4jH7Vj0y3v8kjRzpnQvOo7BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0458230e04d57aa1997f2c21948accc1965dbeb706e9499129f15696d1269661","last_reissued_at":"2026-05-26T01:03:18.313275Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:03:18.313275Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2510.26418","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-26T01:03:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UjnDA8rZ+ERRQSfYGpkrS+XDKWoAwywb6SACH3dxC11fRKqyJAPZRPHkXiS3+A/Svbn6Tw2G55ABgVYf1KGrAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-28T06:16:30.201131Z"},"content_sha256":"50962ca793df145419beb67ae1e3c5522eab986b9ec6a4024007c399fa99de95","schema_version":"1.0","event_id":"sha256:50962ca793df145419beb67ae1e3c5522eab986b9ec6a4024007c399fa99de95"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:ARMCGDQE2V5KDGL7FQQZJCWMYG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Chain-of-Thought Hijacking","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Fazl Barez, Jianli Zhao, Mrinank Sharma, Rylan Schaeffer, Tingchen Fu","submitted_at":"2025-10-30T12:10:03Z","abstract_excerpt":"Large Reasoning Models (LRMs) improve task performance through extended inference-time reasoning. Although previous studies suggest that longer reasoning should lead to more robust safety behavior, we find evidence to the contrary: over-extended reasoning can instead be exploited to systematically weaken refusal behavior. We propose Chain-of-Thought Hijacking, a simple yet effective black-box jailbreak attack that induces LRMs to engage in prolonged benign puzzle-solving reasoning, often lasting more than five minutes, before eliciting harmful compliance. Across HarmBench, CoT Hijacking achiev"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.26418","kind":"arxiv","version":4},"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/2510.26418/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":"2026-05-26T01:03:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rkkr5cqU4HaZtGorKyQYrqtsreZ/GoNvG77F540r+4K1KLByTUp3R5JeFVvK7FE9ux5yFUoMAKQYX5Mls+6cAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-28T06:16:30.201542Z"},"content_sha256":"7eb3b52c1a6be9c8a824ab3fe6a179dd725fc65caf1637a4e03b098a79c8f7a6","schema_version":"1.0","event_id":"sha256:7eb3b52c1a6be9c8a824ab3fe6a179dd725fc65caf1637a4e03b098a79c8f7a6"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG/bundle.json","state_url":"https://pith.science/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG/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":"2026-05-28T06:16:30Z","links":{"resolver":"https://pith.science/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG","bundle":"https://pith.science/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG/bundle.json","state":"https://pith.science/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ARMCGDQE2V5KDGL7FQQZJCWMYG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:ARMCGDQE2V5KDGL7FQQZJCWMYG","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":"5795eb1cda8c62ec4b3e15395dc3f2cccfeb4635e5153e1c1d5f25cc44c6d041","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-10-30T12:10:03Z","title_canon_sha256":"52e93352d29bc904afed2c494aa8028596e71729c2e9167cf9fef3e50293d3b9"},"schema_version":"1.0","source":{"id":"2510.26418","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2510.26418","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"arxiv_version","alias_value":"2510.26418v4","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.26418","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"pith_short_12","alias_value":"ARMCGDQE2V5K","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"pith_short_16","alias_value":"ARMCGDQE2V5KDGL7","created_at":"2026-05-26T01:03:18Z"},{"alias_kind":"pith_short_8","alias_value":"ARMCGDQE","created_at":"2026-05-26T01:03:18Z"}],"graph_snapshots":[{"event_id":"sha256:7eb3b52c1a6be9c8a824ab3fe6a179dd725fc65caf1637a4e03b098a79c8f7a6","target":"graph","created_at":"2026-05-26T01:03:18Z","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/2510.26418/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Reasoning Models (LRMs) improve task performance through extended inference-time reasoning. Although previous studies suggest that longer reasoning should lead to more robust safety behavior, we find evidence to the contrary: over-extended reasoning can instead be exploited to systematically weaken refusal behavior. We propose Chain-of-Thought Hijacking, a simple yet effective black-box jailbreak attack that induces LRMs to engage in prolonged benign puzzle-solving reasoning, often lasting more than five minutes, before eliciting harmful compliance. Across HarmBench, CoT Hijacking achiev","authors_text":"Fazl Barez, Jianli Zhao, Mrinank Sharma, Rylan Schaeffer, Tingchen Fu","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-10-30T12:10:03Z","title":"Chain-of-Thought Hijacking"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.26418","kind":"arxiv","version":4},"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":"sha256:50962ca793df145419beb67ae1e3c5522eab986b9ec6a4024007c399fa99de95","target":"record","created_at":"2026-05-26T01:03:18Z","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":"5795eb1cda8c62ec4b3e15395dc3f2cccfeb4635e5153e1c1d5f25cc44c6d041","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-10-30T12:10:03Z","title_canon_sha256":"52e93352d29bc904afed2c494aa8028596e71729c2e9167cf9fef3e50293d3b9"},"schema_version":"1.0","source":{"id":"2510.26418","kind":"arxiv","version":4}},"canonical_sha256":"0458230e04d57aa1997f2c21948accc1965dbeb706e9499129f15696d1269661","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0458230e04d57aa1997f2c21948accc1965dbeb706e9499129f15696d1269661","first_computed_at":"2026-05-26T01:03:18.313275Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-26T01:03:18.313275Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"NQid23hMHAnrBqQfHR+ILgNrkaYLV+OjSj5UVkEuMBkFQoq+VfVNj28X0/iEwl4jH7Vj0y3v8kjRzpnQvOo7BQ==","signature_status":"signed_v1","signed_at":"2026-05-26T01:03:18.314117Z","signed_message":"canonical_sha256_bytes"},"source_id":"2510.26418","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:50962ca793df145419beb67ae1e3c5522eab986b9ec6a4024007c399fa99de95","sha256:7eb3b52c1a6be9c8a824ab3fe6a179dd725fc65caf1637a4e03b098a79c8f7a6"],"state_sha256":"974c7ce674a80dc5733611f9920c49c031bfec14238c910accb008c67e773d44"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uCDRrqGIoHMPD1wdjbv9AfZ6b5fy71+STasKj7sk1pSppuy7RJf4hR+dQyKAmO57jTu4QFvfIA/ZMIVrSBHUAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-28T06:16:30.203721Z","bundle_sha256":"f9be20080aababfaa27c4ef345747bf13e5a4edadb5d73f94c069a3d896b2e52"}}