{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DBYFOPG4OJ6ZQH3YNXZGGWSKYE","short_pith_number":"pith:DBYFOPG4","schema_version":"1.0","canonical_sha256":"1870573cdc727d981f786df2635a4ac127f875301ae6047ddff09ff9987f492b","source":{"kind":"arxiv","id":"2607.02047","version":1},"attestation_state":"computed","paper":{"title":"OpenSafeIntent: Evaluating Intent-Calibrated Safe Completion Across Dual-Use Prompt Sets","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Junjie Hu, Rheeya Uppaal, Selina Sung, Seungwoo Lyu","submitted_at":"2026-07-02T11:14:52Z","abstract_excerpt":"Safe completion requires models to provide useful assistance without enabling harm, but this behavior is difficult to evaluate with isolated prompts. We introduce OpenSafeIntent, a benchmark of controlled prompt-sets that vary intent while holding the underlying task fixed. Each datapoint contains benign, dual-use, and malicious variants of the same task. This design lets us evaluate whether models calibrate assistance across intent shifts, rather than merely appearing safe on average. Across a broad model suite, we find that prompt-level safety hides important failures: models often fail to r"},"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":"2607.02047","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-02T11:14:52Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1f07026e627544d09b09e1488a36bf9699ad94c0aef670c6c897bc29763b66a2","abstract_canon_sha256":"26e4549c0677f37d20c96fbafd59fc1ee2ce0410ddf5adc8f63935ea82faae1f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-03T01:17:38.038608Z","signature_b64":"VEQVDnRQIuq3lkWVL78C21FIIBJ/2QKI6V/2/smO6beGr42VS4SQPkblzd/Xx0EVXlysWVldH6IWRcXFLhvKAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1870573cdc727d981f786df2635a4ac127f875301ae6047ddff09ff9987f492b","last_reissued_at":"2026-07-03T01:17:38.038158Z","signature_status":"signed_v1","first_computed_at":"2026-07-03T01:17:38.038158Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"OpenSafeIntent: Evaluating Intent-Calibrated Safe Completion Across Dual-Use Prompt Sets","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Junjie Hu, Rheeya Uppaal, Selina Sung, Seungwoo Lyu","submitted_at":"2026-07-02T11:14:52Z","abstract_excerpt":"Safe completion requires models to provide useful assistance without enabling harm, but this behavior is difficult to evaluate with isolated prompts. We introduce OpenSafeIntent, a benchmark of controlled prompt-sets that vary intent while holding the underlying task fixed. Each datapoint contains benign, dual-use, and malicious variants of the same task. This design lets us evaluate whether models calibrate assistance across intent shifts, rather than merely appearing safe on average. Across a broad model suite, we find that prompt-level safety hides important failures: models often fail to r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02047","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/2607.02047/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":"2607.02047","created_at":"2026-07-03T01:17:38.038212+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02047v1","created_at":"2026-07-03T01:17:38.038212+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02047","created_at":"2026-07-03T01:17:38.038212+00:00"},{"alias_kind":"pith_short_12","alias_value":"DBYFOPG4OJ6Z","created_at":"2026-07-03T01:17:38.038212+00:00"},{"alias_kind":"pith_short_16","alias_value":"DBYFOPG4OJ6ZQH3Y","created_at":"2026-07-03T01:17:38.038212+00:00"},{"alias_kind":"pith_short_8","alias_value":"DBYFOPG4","created_at":"2026-07-03T01:17:38.038212+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/DBYFOPG4OJ6ZQH3YNXZGGWSKYE","json":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE.json","graph_json":"https://pith.science/api/pith-number/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/graph.json","events_json":"https://pith.science/api/pith-number/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/events.json","paper":"https://pith.science/paper/DBYFOPG4"},"agent_actions":{"view_html":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE","download_json":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE.json","view_paper":"https://pith.science/paper/DBYFOPG4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02047&json=true","fetch_graph":"https://pith.science/api/pith-number/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/graph.json","fetch_events":"https://pith.science/api/pith-number/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/action/storage_attestation","attest_author":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/action/author_attestation","sign_citation":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/action/citation_signature","submit_replication":"https://pith.science/pith/DBYFOPG4OJ6ZQH3YNXZGGWSKYE/action/replication_record"}},"created_at":"2026-07-03T01:17:38.038212+00:00","updated_at":"2026-07-03T01:17:38.038212+00:00"}