{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:HZETIMLDPKTGCI5F65BIFL6ZWO","short_pith_number":"pith:HZETIMLD","canonical_record":{"source":{"id":"2402.15180","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-23T08:22:24Z","cross_cats_sorted":["cs.CL","cs.CR"],"title_canon_sha256":"d8b58a350241ab74090356f7d03926a795e733f283b1e169f76f4695bd472082","abstract_canon_sha256":"0feb1fb7d0a1002a03a2265482a7faf0aca9d969b59e5249ccf76e13cdee62fe"},"schema_version":"1.0"},"canonical_sha256":"3e493431637aa66123a5f74282afd9b384e0a3d7fc255135b1b10ce222b7bb9e","source":{"kind":"arxiv","id":"2402.15180","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.15180","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"arxiv_version","alias_value":"2402.15180v2","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.15180","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"pith_short_12","alias_value":"HZETIMLDPKTG","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"pith_short_16","alias_value":"HZETIMLDPKTGCI5F","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"pith_short_8","alias_value":"HZETIMLD","created_at":"2026-07-05T07:49:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:HZETIMLDPKTGCI5F65BIFL6ZWO","target":"record","payload":{"canonical_record":{"source":{"id":"2402.15180","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-23T08:22:24Z","cross_cats_sorted":["cs.CL","cs.CR"],"title_canon_sha256":"d8b58a350241ab74090356f7d03926a795e733f283b1e169f76f4695bd472082","abstract_canon_sha256":"0feb1fb7d0a1002a03a2265482a7faf0aca9d969b59e5249ccf76e13cdee62fe"},"schema_version":"1.0"},"canonical_sha256":"3e493431637aa66123a5f74282afd9b384e0a3d7fc255135b1b10ce222b7bb9e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:49:37.583356Z","signature_b64":"tGeKis3blikn/vRpsclzBXveL7gvUJFeQtSIqUN7S6gUIRe5xSyWnaxCXYuGNDPbcyW8DydUVWfCrclV/DzlAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3e493431637aa66123a5f74282afd9b384e0a3d7fc255135b1b10ce222b7bb9e","last_reissued_at":"2026-07-05T07:49:37.582809Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:49:37.582809Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2402.15180","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:49:37Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"i4eGNYkUvgHw7TZZR1rg4TB1nK/OlBb5rhEGclIMlFbWXcGolv0rkiLxjDx8D7GRBEHZjZi6WB//kwmVmfRIBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T19:26:33.702450Z"},"content_sha256":"41a104bab0a4f3acb4d3c117e1622ef5591c75f4ac05a5701328e2eb21f58a00","schema_version":"1.0","event_id":"sha256:41a104bab0a4f3acb4d3c117e1622ef5591c75f4ac05a5701328e2eb21f58a00"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:HZETIMLDPKTGCI5F65BIFL6ZWO","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Break the Breakout: Reinventing LM Defense Against Jailbreak Attacks with Self-Refinement","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Heegyu Kim, Hyunsouk Cho, Sehyun Yuk","submitted_at":"2024-02-23T08:22:24Z","abstract_excerpt":"Caution: This paper includes offensive words that could potentially cause unpleasantness. Language models (LMs) are vulnerable to exploitation for adversarial misuse. Training LMs for safety alignment is extensive and makes it hard to respond to fast-developing attacks immediately, such as jailbreaks. We propose self-refine with formatting that achieves outstanding safety even in non-safety-aligned LMs and evaluate our method alongside several defense baselines, demonstrating that it is the safest training-free method against jailbreak attacks. Additionally, we proposed a formatting method tha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.15180","kind":"arxiv","version":2},"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/2402.15180/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":"2026-07-05T07:49:37Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UfRZi+OPfOg6lW+mzceXHFiWgRhJsRqD0D4mqazRP+oK02+l5YywW4Rg95HZElYPz/KEjTrkH4B6XGJsg/FkAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T19:26:33.703046Z"},"content_sha256":"f1137ca00231bd261965186f0434e64d51737a65e1998980f705fbbd302f8a22","schema_version":"1.0","event_id":"sha256:f1137ca00231bd261965186f0434e64d51737a65e1998980f705fbbd302f8a22"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/HZETIMLDPKTGCI5F65BIFL6ZWO/bundle.json","state_url":"https://pith.science/pith/HZETIMLDPKTGCI5F65BIFL6ZWO/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/HZETIMLDPKTGCI5F65BIFL6ZWO/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":"2026-08-09T19:26:33Z","links":{"resolver":"https://pith.science/pith/HZETIMLDPKTGCI5F65BIFL6ZWO","bundle":"https://pith.science/pith/HZETIMLDPKTGCI5F65BIFL6ZWO/bundle.json","state":"https://pith.science/pith/HZETIMLDPKTGCI5F65BIFL6ZWO/state.json","well_known_bundle":"https://pith.science/.well-known/pith/HZETIMLDPKTGCI5F65BIFL6ZWO/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:HZETIMLDPKTGCI5F65BIFL6ZWO","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":"0feb1fb7d0a1002a03a2265482a7faf0aca9d969b59e5249ccf76e13cdee62fe","cross_cats_sorted":["cs.CL","cs.CR"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-23T08:22:24Z","title_canon_sha256":"d8b58a350241ab74090356f7d03926a795e733f283b1e169f76f4695bd472082"},"schema_version":"1.0","source":{"id":"2402.15180","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.15180","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"arxiv_version","alias_value":"2402.15180v2","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.15180","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"pith_short_12","alias_value":"HZETIMLDPKTG","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"pith_short_16","alias_value":"HZETIMLDPKTGCI5F","created_at":"2026-07-05T07:49:37Z"},{"alias_kind":"pith_short_8","alias_value":"HZETIMLD","created_at":"2026-07-05T07:49:37Z"}],"graph_snapshots":[{"event_id":"sha256:f1137ca00231bd261965186f0434e64d51737a65e1998980f705fbbd302f8a22","target":"graph","created_at":"2026-07-05T07:49:37Z","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/2402.15180/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Caution: This paper includes offensive words that could potentially cause unpleasantness. Language models (LMs) are vulnerable to exploitation for adversarial misuse. Training LMs for safety alignment is extensive and makes it hard to respond to fast-developing attacks immediately, such as jailbreaks. We propose self-refine with formatting that achieves outstanding safety even in non-safety-aligned LMs and evaluate our method alongside several defense baselines, demonstrating that it is the safest training-free method against jailbreak attacks. Additionally, we proposed a formatting method tha","authors_text":"Heegyu Kim, Hyunsouk Cho, Sehyun Yuk","cross_cats":["cs.CL","cs.CR"],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-23T08:22:24Z","title":"Break the Breakout: Reinventing LM Defense Against Jailbreak Attacks with Self-Refinement"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.15180","kind":"arxiv","version":2},"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":"sha256:41a104bab0a4f3acb4d3c117e1622ef5591c75f4ac05a5701328e2eb21f58a00","target":"record","created_at":"2026-07-05T07:49:37Z","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":"0feb1fb7d0a1002a03a2265482a7faf0aca9d969b59e5249ccf76e13cdee62fe","cross_cats_sorted":["cs.CL","cs.CR"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-23T08:22:24Z","title_canon_sha256":"d8b58a350241ab74090356f7d03926a795e733f283b1e169f76f4695bd472082"},"schema_version":"1.0","source":{"id":"2402.15180","kind":"arxiv","version":2}},"canonical_sha256":"3e493431637aa66123a5f74282afd9b384e0a3d7fc255135b1b10ce222b7bb9e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3e493431637aa66123a5f74282afd9b384e0a3d7fc255135b1b10ce222b7bb9e","first_computed_at":"2026-07-05T07:49:37.582809Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:49:37.582809Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"tGeKis3blikn/vRpsclzBXveL7gvUJFeQtSIqUN7S6gUIRe5xSyWnaxCXYuGNDPbcyW8DydUVWfCrclV/DzlAA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:49:37.583356Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.15180","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:41a104bab0a4f3acb4d3c117e1622ef5591c75f4ac05a5701328e2eb21f58a00","sha256:f1137ca00231bd261965186f0434e64d51737a65e1998980f705fbbd302f8a22"],"state_sha256":"50271d53cb2c92920137e91235a97f6c74e4a254aea282c75cacc28889dd75ee"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rwu51X6RGlApsqREd8TxcS7nFiUrpplYUAvWTPmgtD9etujoSd4QTWIytf29Wge1nnRAmIBmu+RHG8SiWKFUAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T19:26:33.706911Z","bundle_sha256":"bb7f3bc6b8ae13ad1c9c249c3de066f680e7520d1d5fbc6512cf96152c6e2375"}}