{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:F5GTU5NN47TFPIUJEKLNXTADMG","short_pith_number":"pith:F5GTU5NN","schema_version":"1.0","canonical_sha256":"2f4d3a75ade7e657a2892296dbcc03619d20955fa4a7e8328d3b1f13e52fd2a5","source":{"kind":"arxiv","id":"2606.03601","version":1},"attestation_state":"computed","paper":{"title":"DDOR: Delta Debugging for Explainable Overrefusal Testing and Repair","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Dongxia Wang, Haonan Zhang, Jun Sun, Peixin Zhang, Qinyan Zhou","submitted_at":"2026-06-02T13:07:12Z","abstract_excerpt":"While safety alignment and guardrails help large language models (LLMs) avoid harmful outputs, they can also induce overrefusal, i.e., unwarranted rejection of benign queries that merely appear risky. We present DDOR (Delta Debugging for OverRefusal), a fully automated and explainable framework for overrefusal testing and repair in a black-box setting, where only model inputs and outputs are accessible and internal safety mechanisms remain opaque. DDOR applies delta debugging to localize minimal refusal-triggering fragments (mRTFs) that provide phrase-level, explainable evidence for why a refu"},"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":"2606.03601","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SE","submitted_at":"2026-06-02T13:07:12Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"532b07cc5142a45d904fd815ed1ad3e907b5fc25f2fc9d3c60cf4e8371ed737d","abstract_canon_sha256":"1556dd409a9dc898d7874028a9070d0888ce592a208e0996f66a335bf0871db7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T01:06:02.046258Z","signature_b64":"/cgDvguaEMmG4aylqFUPsrdjrrSyOQsB8syz6CSYJdPi5mkMYLt59kFYn3edTWpx4WvO1PovJXM8+wNMMO4NAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2f4d3a75ade7e657a2892296dbcc03619d20955fa4a7e8328d3b1f13e52fd2a5","last_reissued_at":"2026-06-03T01:06:02.045779Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T01:06:02.045779Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DDOR: Delta Debugging for Explainable Overrefusal Testing and Repair","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Dongxia Wang, Haonan Zhang, Jun Sun, Peixin Zhang, Qinyan Zhou","submitted_at":"2026-06-02T13:07:12Z","abstract_excerpt":"While safety alignment and guardrails help large language models (LLMs) avoid harmful outputs, they can also induce overrefusal, i.e., unwarranted rejection of benign queries that merely appear risky. We present DDOR (Delta Debugging for OverRefusal), a fully automated and explainable framework for overrefusal testing and repair in a black-box setting, where only model inputs and outputs are accessible and internal safety mechanisms remain opaque. DDOR applies delta debugging to localize minimal refusal-triggering fragments (mRTFs) that provide phrase-level, explainable evidence for why a refu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.03601","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/2606.03601/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":"2606.03601","created_at":"2026-06-03T01:06:02.045855+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.03601v1","created_at":"2026-06-03T01:06:02.045855+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.03601","created_at":"2026-06-03T01:06:02.045855+00:00"},{"alias_kind":"pith_short_12","alias_value":"F5GTU5NN47TF","created_at":"2026-06-03T01:06:02.045855+00:00"},{"alias_kind":"pith_short_16","alias_value":"F5GTU5NN47TFPIUJ","created_at":"2026-06-03T01:06:02.045855+00:00"},{"alias_kind":"pith_short_8","alias_value":"F5GTU5NN","created_at":"2026-06-03T01:06:02.045855+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/F5GTU5NN47TFPIUJEKLNXTADMG","json":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG.json","graph_json":"https://pith.science/api/pith-number/F5GTU5NN47TFPIUJEKLNXTADMG/graph.json","events_json":"https://pith.science/api/pith-number/F5GTU5NN47TFPIUJEKLNXTADMG/events.json","paper":"https://pith.science/paper/F5GTU5NN"},"agent_actions":{"view_html":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG","download_json":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG.json","view_paper":"https://pith.science/paper/F5GTU5NN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.03601&json=true","fetch_graph":"https://pith.science/api/pith-number/F5GTU5NN47TFPIUJEKLNXTADMG/graph.json","fetch_events":"https://pith.science/api/pith-number/F5GTU5NN47TFPIUJEKLNXTADMG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG/action/storage_attestation","attest_author":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG/action/author_attestation","sign_citation":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG/action/citation_signature","submit_replication":"https://pith.science/pith/F5GTU5NN47TFPIUJEKLNXTADMG/action/replication_record"}},"created_at":"2026-06-03T01:06:02.045855+00:00","updated_at":"2026-06-03T01:06:02.045855+00:00"}