{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KIRMQ3MIW456ST6K2EZLVYVPQB","short_pith_number":"pith:KIRMQ3MI","schema_version":"1.0","canonical_sha256":"5222c86d88b73be94fcad132bae2af805ffc1ee215442e77efb03b9bbe32ea38","source":{"kind":"arxiv","id":"2607.16442","version":1},"attestation_state":"computed","paper":{"title":"One Modality to Forget Them All: Enhancing Cross-Modal Unlearning in Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.CV","authors_text":"Guangjing Wang, Ning Wang, Sudharshan Balaji, Yili Ren, Yimin Chen","submitted_at":"2026-07-17T18:44:02Z","abstract_excerpt":"Machine unlearning is widely used to remove hazardous knowledge from large language models. Modern Vision-Language Models (VLMs), however, process both text and visual inputs, raising a fundamental security question: does unlearning in one modality transfer to the other? We present the first systematic, bidirectional study of cross-modal unlearning transfer across three VLM architectures: LLaVA-1.5 (MLP projection), InstructBLIP (Q-Former), and IDEFICS (gated cross-attention). We find that unlearning transfers across modalities, but the transfer is asymmetric and incomplete. In some cases, tex"},"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":"2607.16442","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-07-17T18:44:02Z","cross_cats_sorted":["cs.CL","cs.CR"],"title_canon_sha256":"a2d1b7dca10de364e2f6390aa310d88afb9a5386e9a0fc446512fd208ce41324","abstract_canon_sha256":"9c00064429b0fb58e5f757eb43349dd9ee0cd595a604f0649a72d146eeb421a9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T00:20:18.868328Z","signature_b64":"x9KUczjI6xcFW1dy/xL0tGRK47F22ntKA/EK8SIX809FMCkCSG32ozQiz3TotYSE2BUD9QwvLEpd1ejTM9icAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5222c86d88b73be94fcad132bae2af805ffc1ee215442e77efb03b9bbe32ea38","last_reissued_at":"2026-07-21T00:20:18.867484Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T00:20:18.867484Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"One Modality to Forget Them All: Enhancing Cross-Modal Unlearning in Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.CV","authors_text":"Guangjing Wang, Ning Wang, Sudharshan Balaji, Yili Ren, Yimin Chen","submitted_at":"2026-07-17T18:44:02Z","abstract_excerpt":"Machine unlearning is widely used to remove hazardous knowledge from large language models. Modern Vision-Language Models (VLMs), however, process both text and visual inputs, raising a fundamental security question: does unlearning in one modality transfer to the other? We present the first systematic, bidirectional study of cross-modal unlearning transfer across three VLM architectures: LLaVA-1.5 (MLP projection), InstructBLIP (Q-Former), and IDEFICS (gated cross-attention). We find that unlearning transfers across modalities, but the transfer is asymmetric and incomplete. In some cases, tex"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.16442","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/2607.16442/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":"2607.16442","created_at":"2026-07-21T00:20:18.867908+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.16442v1","created_at":"2026-07-21T00:20:18.867908+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.16442","created_at":"2026-07-21T00:20:18.867908+00:00"},{"alias_kind":"pith_short_12","alias_value":"KIRMQ3MIW456","created_at":"2026-07-21T00:20:18.867908+00:00"},{"alias_kind":"pith_short_16","alias_value":"KIRMQ3MIW456ST6K","created_at":"2026-07-21T00:20:18.867908+00:00"},{"alias_kind":"pith_short_8","alias_value":"KIRMQ3MI","created_at":"2026-07-21T00:20:18.867908+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/KIRMQ3MIW456ST6K2EZLVYVPQB","json":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB.json","graph_json":"https://pith.science/api/pith-number/KIRMQ3MIW456ST6K2EZLVYVPQB/graph.json","events_json":"https://pith.science/api/pith-number/KIRMQ3MIW456ST6K2EZLVYVPQB/events.json","paper":"https://pith.science/paper/KIRMQ3MI"},"agent_actions":{"view_html":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB","download_json":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB.json","view_paper":"https://pith.science/paper/KIRMQ3MI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.16442&json=true","fetch_graph":"https://pith.science/api/pith-number/KIRMQ3MIW456ST6K2EZLVYVPQB/graph.json","fetch_events":"https://pith.science/api/pith-number/KIRMQ3MIW456ST6K2EZLVYVPQB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB/action/storage_attestation","attest_author":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB/action/author_attestation","sign_citation":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB/action/citation_signature","submit_replication":"https://pith.science/pith/KIRMQ3MIW456ST6K2EZLVYVPQB/action/replication_record"}},"created_at":"2026-07-21T00:20:18.867908+00:00","updated_at":"2026-07-21T00:20:18.867908+00:00"}