{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:PSAKYGRBQPSN2B36P6WB4HGECT","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":"69de6bd98f3dce646256f7dfdb6cc3de53c757c93b81c643f1427bbcaacbf53d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-25T07:15:23Z","title_canon_sha256":"448a3b5963deec416eb2eed57b5cf5e49e2175e4ede0d29e4f0f57bcd35a3f69"},"schema_version":"1.0","source":{"id":"2404.16369","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.16369","created_at":"2026-07-05T11:30:35Z"},{"alias_kind":"arxiv_version","alias_value":"2404.16369v3","created_at":"2026-07-05T11:30:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.16369","created_at":"2026-07-05T11:30:35Z"},{"alias_kind":"pith_short_12","alias_value":"PSAKYGRBQPSN","created_at":"2026-07-05T11:30:35Z"},{"alias_kind":"pith_short_16","alias_value":"PSAKYGRBQPSN2B36","created_at":"2026-07-05T11:30:35Z"},{"alias_kind":"pith_short_8","alias_value":"PSAKYGRB","created_at":"2026-07-05T11:30:35Z"}],"graph_snapshots":[{"event_id":"sha256:c4e3218b9e7b71ff72cec1e97507705423590c57ab022fa8307a38ed37b1474e","target":"graph","created_at":"2026-07-05T11:30:35Z","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/2404.16369/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Ensuring the safety alignment of Large Language Models (LLMs) is critical for generating responses consistent with human values. However, LLMs remain vulnerable to jailbreaking attacks, where carefully crafted prompts manipulate them into producing toxic content. One category of such attacks reformulates the task as an optimization problem, aiming to elicit affirmative responses from the LLM. However, these methods heavily rely on predefined objectionable behaviors, limiting their effectiveness and adaptability to diverse harmful queries. In this study, we first identify why the vanilla target","authors_text":"Jian Lou, Wenjie Wang, Yibei Yang, Yukai Zhou, Zhan Qin, Zhijie Huang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-25T07:15:23Z","title":"Don't Say No: Jailbreaking LLM by Suppressing Refusal"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.16369","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:c7247ec3b7e319ce63d90aa93d2e51371044e6fd638e76a7aa69f6237bf4033f","target":"record","created_at":"2026-07-05T11:30:35Z","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":"69de6bd98f3dce646256f7dfdb6cc3de53c757c93b81c643f1427bbcaacbf53d","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-25T07:15:23Z","title_canon_sha256":"448a3b5963deec416eb2eed57b5cf5e49e2175e4ede0d29e4f0f57bcd35a3f69"},"schema_version":"1.0","source":{"id":"2404.16369","kind":"arxiv","version":3}},"canonical_sha256":"7c80ac1a2183e4dd077e7fac1e1cc414d07a1b404eb807288f2dd3985b6811a2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"7c80ac1a2183e4dd077e7fac1e1cc414d07a1b404eb807288f2dd3985b6811a2","first_computed_at":"2026-07-05T11:30:35.886739Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:30:35.886739Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"oJmnH4itnzKx6IaQEQnOYjz7yLpcXCIzMNniBDLIdfJ4u50R1T7huHEoCgHyEQxO4Bx7BRb2pQrn3HDsOYomCg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:30:35.887240Z","signed_message":"canonical_sha256_bytes"},"source_id":"2404.16369","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c7247ec3b7e319ce63d90aa93d2e51371044e6fd638e76a7aa69f6237bf4033f","sha256:c4e3218b9e7b71ff72cec1e97507705423590c57ab022fa8307a38ed37b1474e"],"state_sha256":"9f4a3ed84a6e214853e99acbbde89dd1f1f996fa1a4e9c4323f0c8bf45ec14ab"}