{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:AFZSKR6WJWHFPXP3HUGRJP3RK6","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":"71b3b77caba9b97e68addd0b23aadb9dea39863386901a40737b0f4a6a601d7e","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T18:48:37Z","title_canon_sha256":"f1fdf0f3b1a8b97bb433da7a8e61f55d094e0417fd62595a3c2d628098e0de3e"},"schema_version":"1.0","source":{"id":"2401.17256","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2401.17256","created_at":"2026-07-05T11:42:20Z"},{"alias_kind":"arxiv_version","alias_value":"2401.17256v5","created_at":"2026-07-05T11:42:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.17256","created_at":"2026-07-05T11:42:20Z"},{"alias_kind":"pith_short_12","alias_value":"AFZSKR6WJWHF","created_at":"2026-07-05T11:42:20Z"},{"alias_kind":"pith_short_16","alias_value":"AFZSKR6WJWHFPXP3","created_at":"2026-07-05T11:42:20Z"},{"alias_kind":"pith_short_8","alias_value":"AFZSKR6W","created_at":"2026-07-05T11:42:20Z"}],"graph_snapshots":[{"event_id":"sha256:ae12b2e50446769f24cf10e7698ff63dfdafc20ba15bdc98a943b157ba5303fc","target":"graph","created_at":"2026-07-05T11:42:20Z","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/2401.17256/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are vulnerable to jailbreak attacks - resulting in harmful, unethical, or biased text generations. However, existing jailbreaking methods are computationally costly. In this paper, we propose the weak-to-strong jailbreaking attack, an efficient inference time attack for aligned LLMs to produce harmful text. Our key intuition is based on the observation that jailbroken and aligned models only differ in their initial decoding distributions. The weak-to-strong attack's key technical insight is using two smaller models (a safe and an unsafe one) to adversarially modify","authors_text":"Chao Du, Lei Li, Tianyu Pang, William Yang Wang, Xianjun Yang, Xuandong Zhao, Yu-Xiang Wang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T18:48:37Z","title":"Weak-to-Strong Jailbreaking on Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.17256","kind":"arxiv","version":5},"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:bd304bd7bb17fd7778093870cc4d2a8e9e414c2e3348200b7b589a0d116c2a1a","target":"record","created_at":"2026-07-05T11:42:20Z","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":"71b3b77caba9b97e68addd0b23aadb9dea39863386901a40737b0f4a6a601d7e","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T18:48:37Z","title_canon_sha256":"f1fdf0f3b1a8b97bb433da7a8e61f55d094e0417fd62595a3c2d628098e0de3e"},"schema_version":"1.0","source":{"id":"2401.17256","kind":"arxiv","version":5}},"canonical_sha256":"01732547d64d8e57ddfb3d0d14bf7157a13c8cf1323f3d36f66a259b32498764","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"01732547d64d8e57ddfb3d0d14bf7157a13c8cf1323f3d36f66a259b32498764","first_computed_at":"2026-07-05T11:42:20.252322Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:42:20.252322Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"i+QotwRU4YxatnKAjb10BYD8nk0GlY4p7yj1D5I1l4scfA7VueIpwMBblBD4RkBigKqv8hzMxDnQJ5d1YD+WAg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:42:20.252722Z","signed_message":"canonical_sha256_bytes"},"source_id":"2401.17256","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bd304bd7bb17fd7778093870cc4d2a8e9e414c2e3348200b7b589a0d116c2a1a","sha256:ae12b2e50446769f24cf10e7698ff63dfdafc20ba15bdc98a943b157ba5303fc"],"state_sha256":"1f155d20cf124890ce23aa4e44fa4c1c572519d42fb4912283ab7a3af7226873"}