{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:XBMMC3LRUBL5YOCHWL3HDLDE3Q","short_pith_number":"pith:XBMMC3LR","canonical_record":{"source":{"id":"2405.13077","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-05-21T03:16:35Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"9de901d27036187cff3e665918f31b731939d94663259768055b5bfb08f73e69","abstract_canon_sha256":"8daa7ae132bb987725a16192dfe862aba75ae032faf8e85e94797e9f6145d638"},"schema_version":"1.0"},"canonical_sha256":"b858c16d71a057dc3847b2f671ac64dc10538f17a5df7f9c8560888b6806c9ab","source":{"kind":"arxiv","id":"2405.13077","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.13077","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"arxiv_version","alias_value":"2405.13077v2","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.13077","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"pith_short_12","alias_value":"XBMMC3LRUBL5","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"pith_short_16","alias_value":"XBMMC3LRUBL5YOCH","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"pith_short_8","alias_value":"XBMMC3LR","created_at":"2026-07-05T09:21:04Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:XBMMC3LRUBL5YOCHWL3HDLDE3Q","target":"record","payload":{"canonical_record":{"source":{"id":"2405.13077","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-05-21T03:16:35Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"9de901d27036187cff3e665918f31b731939d94663259768055b5bfb08f73e69","abstract_canon_sha256":"8daa7ae132bb987725a16192dfe862aba75ae032faf8e85e94797e9f6145d638"},"schema_version":"1.0"},"canonical_sha256":"b858c16d71a057dc3847b2f671ac64dc10538f17a5df7f9c8560888b6806c9ab","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:21:04.737866Z","signature_b64":"DWpyXGx6jH0/WAEftCwwXFNA7clRT2H4tns37jRyvbMtD5WhtuAmV4C+yeK7evxNvE9wp746KKXDNfigfWhLAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b858c16d71a057dc3847b2f671ac64dc10538f17a5df7f9c8560888b6806c9ab","last_reissued_at":"2026-07-05T09:21:04.737344Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:21:04.737344Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.13077","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":"2026-07-05T09:21:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"r61XGjvblCtjPD607qbl/x0pN8eYgSBLy/6v8y5SQNRJY53O7stl+HhyWJroYVq9Kc2IU9HpR0K5yB94Mq+zBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T21:26:35.920812Z"},"content_sha256":"04f4a5eb959cfa5b3b487f964d395a9c9844f284e7fc9b96e6fffe8f2aca4790","schema_version":"1.0","event_id":"sha256:04f4a5eb959cfa5b3b487f964d395a9c9844f284e7fc9b96e6fffe8f2aca4790"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:XBMMC3LRUBL5YOCHWL3HDLDE3Q","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"GPT-4 Jailbreaks Itself with Near-Perfect Success Using Self-Explanation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CR","authors_text":"Govind Ramesh, Wei Xu, Yao Dou","submitted_at":"2024-05-21T03:16:35Z","abstract_excerpt":"Research on jailbreaking has been valuable for testing and understanding the safety and security issues of large language models (LLMs). In this paper, we introduce Iterative Refinement Induced Self-Jailbreak (IRIS), a novel approach that leverages the reflective capabilities of LLMs for jailbreaking with only black-box access. Unlike previous methods, IRIS simplifies the jailbreaking process by using a single model as both the attacker and target. This method first iteratively refines adversarial prompts through self-explanation, which is crucial for ensuring that even well-aligned LLMs obey "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.13077","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/2405.13077/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":"2026-07-05T09:21:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4wu6oYVgl1q0r0R0aqRBniNuxgGnO8I7/SvN14iSu0LBN3OZoGVcIhl5lzA0YknR3p/vpmEF2s5OCcIioXx8Cw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T21:26:35.921198Z"},"content_sha256":"b949517d38f6d03e1c1ab8da29d2b2457945e06071568bd84f445110845d7c35","schema_version":"1.0","event_id":"sha256:b949517d38f6d03e1c1ab8da29d2b2457945e06071568bd84f445110845d7c35"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q/bundle.json","state_url":"https://pith.science/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q/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":"2026-08-14T21:26:35Z","links":{"resolver":"https://pith.science/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q","bundle":"https://pith.science/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q/bundle.json","state":"https://pith.science/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XBMMC3LRUBL5YOCHWL3HDLDE3Q/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:XBMMC3LRUBL5YOCHWL3HDLDE3Q","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":"8daa7ae132bb987725a16192dfe862aba75ae032faf8e85e94797e9f6145d638","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-05-21T03:16:35Z","title_canon_sha256":"9de901d27036187cff3e665918f31b731939d94663259768055b5bfb08f73e69"},"schema_version":"1.0","source":{"id":"2405.13077","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.13077","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"arxiv_version","alias_value":"2405.13077v2","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.13077","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"pith_short_12","alias_value":"XBMMC3LRUBL5","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"pith_short_16","alias_value":"XBMMC3LRUBL5YOCH","created_at":"2026-07-05T09:21:04Z"},{"alias_kind":"pith_short_8","alias_value":"XBMMC3LR","created_at":"2026-07-05T09:21:04Z"}],"graph_snapshots":[{"event_id":"sha256:b949517d38f6d03e1c1ab8da29d2b2457945e06071568bd84f445110845d7c35","target":"graph","created_at":"2026-07-05T09:21:04Z","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/2405.13077/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Research on jailbreaking has been valuable for testing and understanding the safety and security issues of large language models (LLMs). In this paper, we introduce Iterative Refinement Induced Self-Jailbreak (IRIS), a novel approach that leverages the reflective capabilities of LLMs for jailbreaking with only black-box access. Unlike previous methods, IRIS simplifies the jailbreaking process by using a single model as both the attacker and target. This method first iteratively refines adversarial prompts through self-explanation, which is crucial for ensuring that even well-aligned LLMs obey ","authors_text":"Govind Ramesh, Wei Xu, Yao Dou","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-05-21T03:16:35Z","title":"GPT-4 Jailbreaks Itself with Near-Perfect Success Using Self-Explanation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.13077","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":"sha256:04f4a5eb959cfa5b3b487f964d395a9c9844f284e7fc9b96e6fffe8f2aca4790","target":"record","created_at":"2026-07-05T09:21:04Z","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":"8daa7ae132bb987725a16192dfe862aba75ae032faf8e85e94797e9f6145d638","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-05-21T03:16:35Z","title_canon_sha256":"9de901d27036187cff3e665918f31b731939d94663259768055b5bfb08f73e69"},"schema_version":"1.0","source":{"id":"2405.13077","kind":"arxiv","version":2}},"canonical_sha256":"b858c16d71a057dc3847b2f671ac64dc10538f17a5df7f9c8560888b6806c9ab","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b858c16d71a057dc3847b2f671ac64dc10538f17a5df7f9c8560888b6806c9ab","first_computed_at":"2026-07-05T09:21:04.737344Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:21:04.737344Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"DWpyXGx6jH0/WAEftCwwXFNA7clRT2H4tns37jRyvbMtD5WhtuAmV4C+yeK7evxNvE9wp746KKXDNfigfWhLAw==","signature_status":"signed_v1","signed_at":"2026-07-05T09:21:04.737866Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.13077","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:04f4a5eb959cfa5b3b487f964d395a9c9844f284e7fc9b96e6fffe8f2aca4790","sha256:b949517d38f6d03e1c1ab8da29d2b2457945e06071568bd84f445110845d7c35"],"state_sha256":"3602f2c0117862aa70a1d5032f13425fb732df928e2ee5b89ed9aa6e347f0f8a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Xq70ZXymx8YHl2JBUklU4Xz27lm+ztVnjvXH8CNgoBV+2Lfcgurk4+JDMGsnnwiif9bldf/h1P8DE8sSfrxfBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-14T21:26:35.923668Z","bundle_sha256":"28395606578948fdab78847f26a2ff080fc52f82faa99f47b5dcbaa291da4f86"}}