{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:PMTSRNFZHY7ZY6EF2G4SECS2LF","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":"4b1522d48a4c019bfb4aff6a2ad7b6602bcb7d3503c7d32fdac1bc23daceb47b","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-05-22T08:22:57Z","title_canon_sha256":"6c2c9198ac2af62bfea04fbff76c5bdcf923973e3bcea206782edef89ba109d9"},"schema_version":"1.0","source":{"id":"2505.17147","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.17147","created_at":"2026-07-05T11:07:47Z"},{"alias_kind":"arxiv_version","alias_value":"2505.17147v1","created_at":"2026-07-05T11:07:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17147","created_at":"2026-07-05T11:07:47Z"},{"alias_kind":"pith_short_12","alias_value":"PMTSRNFZHY7Z","created_at":"2026-07-05T11:07:47Z"},{"alias_kind":"pith_short_16","alias_value":"PMTSRNFZHY7ZY6EF","created_at":"2026-07-05T11:07:47Z"},{"alias_kind":"pith_short_8","alias_value":"PMTSRNFZ","created_at":"2026-07-05T11:07:47Z"}],"graph_snapshots":[{"event_id":"sha256:58338a5ec9dc3fd9522db19ba7cf5fe28e090e735f6fba713bce5544cea3892d","target":"graph","created_at":"2026-07-05T11:07:47Z","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/2505.17147/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The proliferation of jailbreak attacks against large language models (LLMs) highlights the need for robust security measures. However, in multi-round dialogues, malicious intentions may be hidden in interactions, leading LLMs to be more prone to produce harmful responses. In this paper, we propose the \\textbf{M}ulti-\\textbf{T}urn \\textbf{S}afety \\textbf{A}lignment (\\ourapproach) framework, to address the challenge of securing LLMs in multi-round interactions. It consists of two stages: In the thought-guided attack learning stage, the red-team model learns about thought-guided multi-round jailb","authors_text":"Daojing He, Jing Li, Jun Yu, Min Zhang, Weiyang Guo, Wenya Wang, Yu Li","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-05-22T08:22:57Z","title":"MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17147","kind":"arxiv","version":1},"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":"sha256:25062b1af8a478d47e11ab48870917cf3cc0dce0d00feab5164f905d459535a9","target":"record","created_at":"2026-07-05T11:07:47Z","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":"4b1522d48a4c019bfb4aff6a2ad7b6602bcb7d3503c7d32fdac1bc23daceb47b","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-05-22T08:22:57Z","title_canon_sha256":"6c2c9198ac2af62bfea04fbff76c5bdcf923973e3bcea206782edef89ba109d9"},"schema_version":"1.0","source":{"id":"2505.17147","kind":"arxiv","version":1}},"canonical_sha256":"7b2728b4b93e3f9c7885d1b9220a5a594fbc854525d5f79450e9b21698dfed56","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"7b2728b4b93e3f9c7885d1b9220a5a594fbc854525d5f79450e9b21698dfed56","first_computed_at":"2026-07-05T11:07:47.830794Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:07:47.830794Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Iug/tF0tnCi9Ky2bPxb2Knth/gP224sEswXJvKh30k1HF6xedC9znTc43W+0BREDSvycMUY8mCVOYBIOabvPDQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:07:47.831273Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.17147","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:25062b1af8a478d47e11ab48870917cf3cc0dce0d00feab5164f905d459535a9","sha256:58338a5ec9dc3fd9522db19ba7cf5fe28e090e735f6fba713bce5544cea3892d"],"state_sha256":"75446505b2c8a769292d56933c659e4618a886eb110c7b41f0cb30558dba400a"}