{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:L7KN7QDB6DQMV33AGPUTYUEC62","short_pith_number":"pith:L7KN7QDB","schema_version":"1.0","canonical_sha256":"5fd4dfc061f0e0caef6033e93c5082f6aedaaf51d579be6b7a3bd06fa7a38886","source":{"kind":"arxiv","id":"2607.00402","version":1},"attestation_state":"computed","paper":{"title":"The Illusion of High Utility in Safety Alignment of Text-to-Image Diffusion Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Adeel Yousaf, Amrit Singh Bedi, James Beetham, Mubarak Shah, Soumik Ghosh","submitted_at":"2026-07-01T04:00:27Z","abstract_excerpt":"Safety alignment of text-to-image (T2I) diffusion models aims to suppress harmful generations while preserving utility on benign prompts. Recent methods often appear to deliver high safety with high utility, but this conclusion rests largely on coarse global utility metrics (e.g., FID, CLIPScore) that are insensitive to fine-grained semantic correctness, creating an illusion of high utility. We show that when utility is measured with structured evaluation, this illusion breaks: on TIFA (Text-to-Image Faithfulness evaluation with Question Answering), safety-aligned models suffer substantial dro"},"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":"2607.00402","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-07-01T04:00:27Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"1132d36d3a3ad461b3a4a6ba7c5a2f6a21dec1cbe197b86042908288b5f6697d","abstract_canon_sha256":"ce6c0648847214299534ca420bad23ba2104e9335d51c1398f4de28a798e3980"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T01:17:42.388632Z","signature_b64":"AJyAV5xIopBaWfJp5mi1jHvIYJE8FBRkW2JTkCzlTJV9HUvMzeGwp6z5IzyrIAigjURo+KLXAlVwdcJYd0k6DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5fd4dfc061f0e0caef6033e93c5082f6aedaaf51d579be6b7a3bd06fa7a38886","last_reissued_at":"2026-07-02T01:17:42.388278Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T01:17:42.388278Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Illusion of High Utility in Safety Alignment of Text-to-Image Diffusion Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Adeel Yousaf, Amrit Singh Bedi, James Beetham, Mubarak Shah, Soumik Ghosh","submitted_at":"2026-07-01T04:00:27Z","abstract_excerpt":"Safety alignment of text-to-image (T2I) diffusion models aims to suppress harmful generations while preserving utility on benign prompts. Recent methods often appear to deliver high safety with high utility, but this conclusion rests largely on coarse global utility metrics (e.g., FID, CLIPScore) that are insensitive to fine-grained semantic correctness, creating an illusion of high utility. We show that when utility is measured with structured evaluation, this illusion breaks: on TIFA (Text-to-Image Faithfulness evaluation with Question Answering), safety-aligned models suffer substantial dro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00402","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/2607.00402/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":"2607.00402","created_at":"2026-07-02T01:17:42.388340+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.00402v1","created_at":"2026-07-02T01:17:42.388340+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00402","created_at":"2026-07-02T01:17:42.388340+00:00"},{"alias_kind":"pith_short_12","alias_value":"L7KN7QDB6DQM","created_at":"2026-07-02T01:17:42.388340+00:00"},{"alias_kind":"pith_short_16","alias_value":"L7KN7QDB6DQMV33A","created_at":"2026-07-02T01:17:42.388340+00:00"},{"alias_kind":"pith_short_8","alias_value":"L7KN7QDB","created_at":"2026-07-02T01:17:42.388340+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/L7KN7QDB6DQMV33AGPUTYUEC62","json":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62.json","graph_json":"https://pith.science/api/pith-number/L7KN7QDB6DQMV33AGPUTYUEC62/graph.json","events_json":"https://pith.science/api/pith-number/L7KN7QDB6DQMV33AGPUTYUEC62/events.json","paper":"https://pith.science/paper/L7KN7QDB"},"agent_actions":{"view_html":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62","download_json":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62.json","view_paper":"https://pith.science/paper/L7KN7QDB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.00402&json=true","fetch_graph":"https://pith.science/api/pith-number/L7KN7QDB6DQMV33AGPUTYUEC62/graph.json","fetch_events":"https://pith.science/api/pith-number/L7KN7QDB6DQMV33AGPUTYUEC62/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62/action/storage_attestation","attest_author":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62/action/author_attestation","sign_citation":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62/action/citation_signature","submit_replication":"https://pith.science/pith/L7KN7QDB6DQMV33AGPUTYUEC62/action/replication_record"}},"created_at":"2026-07-02T01:17:42.388340+00:00","updated_at":"2026-07-02T01:17:42.388340+00:00"}