{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:S6W53TJI7LHMPDQ3TQG3EJDM6K","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":"09e21f4b825bf71966d402c50cc7cfa2493e450af99cd46e1bb93808379f5972","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-11-10T13:51:51Z","title_canon_sha256":"27277c2bdb3530cdb7af0fe5de17cb9118066605ae6699baffc3733154439e8f"},"schema_version":"1.0","source":{"id":"2511.07107","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2511.07107","created_at":"2026-06-04T01:09:39Z"},{"alias_kind":"arxiv_version","alias_value":"2511.07107v3","created_at":"2026-06-04T01:09:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.07107","created_at":"2026-06-04T01:09:39Z"},{"alias_kind":"pith_short_12","alias_value":"S6W53TJI7LHM","created_at":"2026-06-04T01:09:39Z"},{"alias_kind":"pith_short_16","alias_value":"S6W53TJI7LHMPDQ3","created_at":"2026-06-04T01:09:39Z"},{"alias_kind":"pith_short_8","alias_value":"S6W53TJI","created_at":"2026-06-04T01:09:39Z"}],"graph_snapshots":[{"event_id":"sha256:a8d528f293f91e4cb71c99741c7e6e265a76f61054f5583ff5a6863eb9b7b3ec","target":"graph","created_at":"2026-06-04T01:09:39Z","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/2511.07107/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Ensuring the safety of Large Language Models (LLMs) is critical for real-world deployment. However, current safety measures often fail to address implicit, domain-specific risks. To investigate this gap, we introduce a dataset of 3,000 annotated queries spanning education, finance, and management. Evaluations across 14 leading LLMs reveal a concerning vulnerability: an average jailbreak success rate of 57.8\\%. In response, we propose MENTOR, a metacognition-driven self-evolution framework. MENTOR performs metacognitive self-assessment, using strategies such as perspective-taking and consequent","authors_text":"Chaochao Lu, Guangze Ye, Guoqing Wang, Jie Zhou, Jingqi Huang, Kaicheng Shen, Liang He, Liang Shan, Qingshan Liu, Wen Wu, Yan Teng, Zhenyu Ying","cross_cats":["cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-11-10T13:51:51Z","title":"MENTOR: A Metacognition-Driven Self-Evolution Framework for Uncovering and Mitigating Implicit Domain Risks in LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.07107","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":"sha256:3711eb9a35cbc48823bf629ca98753e2c42d9fbca9883eab71de07cfebc49307","target":"record","created_at":"2026-06-04T01:09:39Z","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":"09e21f4b825bf71966d402c50cc7cfa2493e450af99cd46e1bb93808379f5972","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-11-10T13:51:51Z","title_canon_sha256":"27277c2bdb3530cdb7af0fe5de17cb9118066605ae6699baffc3733154439e8f"},"schema_version":"1.0","source":{"id":"2511.07107","kind":"arxiv","version":3}},"canonical_sha256":"97adddcd28facec78e1b9c0db2246cf28a256feade60983761c901fecfd742fa","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"97adddcd28facec78e1b9c0db2246cf28a256feade60983761c901fecfd742fa","first_computed_at":"2026-06-04T01:09:39.411610Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-04T01:09:39.411610Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"RMthFigLbxdtKsFQ31pHkV5wAvORn02Y2HPuLhOFyatoWHhQ6UaLVE6yA9FoRZL7xzSeKBkmNhSM1xIYcLA2Ag==","signature_status":"signed_v1","signed_at":"2026-06-04T01:09:39.412187Z","signed_message":"canonical_sha256_bytes"},"source_id":"2511.07107","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3711eb9a35cbc48823bf629ca98753e2c42d9fbca9883eab71de07cfebc49307","sha256:a8d528f293f91e4cb71c99741c7e6e265a76f61054f5583ff5a6863eb9b7b3ec"],"state_sha256":"de533fdf37abd2aa42ec8b38484c5ba9ff1750fe1c68a98a5719f1a1852f71b0"}