{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YR5PJWHLMFIOV5ETQQHOBTFCCN","short_pith_number":"pith:YR5PJWHL","schema_version":"1.0","canonical_sha256":"c47af4d8eb6150eaf493840ee0cca21360d6b9db6e47ce3e19082d09e7371c78","source":{"kind":"arxiv","id":"2505.12060","version":1},"attestation_state":"computed","paper":{"title":"Why Not Act on What You Know? Unleashing Safety Potential of LLMs via Self-Aware Guard Enhancement","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiajun Chen, Jun Kuang, Peng Ding, Shujian Huang, Xuezhi Cao, Xunliang Cai, Zongyu Wang","submitted_at":"2025-05-17T15:54:52Z","abstract_excerpt":"Large Language Models (LLMs) have shown impressive capabilities across various tasks but remain vulnerable to meticulously crafted jailbreak attacks. In this paper, we identify a critical safety gap: while LLMs are adept at detecting jailbreak prompts, they often produce unsafe responses when directly processing these inputs. Inspired by this insight, we propose SAGE (Self-Aware Guard Enhancement), a training-free defense strategy designed to align LLMs' strong safety discrimination performance with their relatively weaker safety generation ability. SAGE consists of two core components: a Disc"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2505.12060","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-17T15:54:52Z","cross_cats_sorted":[],"title_canon_sha256":"3ef38260bfb8a44af3f7dd0fa6607d64b91d8238a8db40b4648a969c21b0b26e","abstract_canon_sha256":"bd2fda9bb647d1336311159b1fd9751160fbe43c3d5bccc79b637b4541372ba7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:04:57.817630Z","signature_b64":"hc8ScTA33J6+Lemj/K34WuA6brgd5UZhkFE0pTJAjh4ecah5eEf3xKoLekK/oAmKbeRX6yaM2QNOKCO+PWPpDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c47af4d8eb6150eaf493840ee0cca21360d6b9db6e47ce3e19082d09e7371c78","last_reissued_at":"2026-07-05T11:04:57.817147Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:04:57.817147Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Why Not Act on What You Know? Unleashing Safety Potential of LLMs via Self-Aware Guard Enhancement","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiajun Chen, Jun Kuang, Peng Ding, Shujian Huang, Xuezhi Cao, Xunliang Cai, Zongyu Wang","submitted_at":"2025-05-17T15:54:52Z","abstract_excerpt":"Large Language Models (LLMs) have shown impressive capabilities across various tasks but remain vulnerable to meticulously crafted jailbreak attacks. In this paper, we identify a critical safety gap: while LLMs are adept at detecting jailbreak prompts, they often produce unsafe responses when directly processing these inputs. Inspired by this insight, we propose SAGE (Self-Aware Guard Enhancement), a training-free defense strategy designed to align LLMs' strong safety discrimination performance with their relatively weaker safety generation ability. SAGE consists of two core components: a Disc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.12060","kind":"arxiv","version":1},"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/2505.12060/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2505.12060","created_at":"2026-07-05T11:04:57.817212+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.12060v1","created_at":"2026-07-05T11:04:57.817212+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.12060","created_at":"2026-07-05T11:04:57.817212+00:00"},{"alias_kind":"pith_short_12","alias_value":"YR5PJWHLMFIO","created_at":"2026-07-05T11:04:57.817212+00:00"},{"alias_kind":"pith_short_16","alias_value":"YR5PJWHLMFIOV5ET","created_at":"2026-07-05T11:04:57.817212+00:00"},{"alias_kind":"pith_short_8","alias_value":"YR5PJWHL","created_at":"2026-07-05T11:04:57.817212+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN","json":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN.json","graph_json":"https://pith.science/api/pith-number/YR5PJWHLMFIOV5ETQQHOBTFCCN/graph.json","events_json":"https://pith.science/api/pith-number/YR5PJWHLMFIOV5ETQQHOBTFCCN/events.json","paper":"https://pith.science/paper/YR5PJWHL"},"agent_actions":{"view_html":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN","download_json":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN.json","view_paper":"https://pith.science/paper/YR5PJWHL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.12060&json=true","fetch_graph":"https://pith.science/api/pith-number/YR5PJWHLMFIOV5ETQQHOBTFCCN/graph.json","fetch_events":"https://pith.science/api/pith-number/YR5PJWHLMFIOV5ETQQHOBTFCCN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN/action/storage_attestation","attest_author":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN/action/author_attestation","sign_citation":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN/action/citation_signature","submit_replication":"https://pith.science/pith/YR5PJWHLMFIOV5ETQQHOBTFCCN/action/replication_record"}},"created_at":"2026-07-05T11:04:57.817212+00:00","updated_at":"2026-07-05T11:04:57.817212+00:00"}