{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:H3PYFHE5QLLVDDPSHCJ6ALBEID","short_pith_number":"pith:H3PYFHE5","schema_version":"1.0","canonical_sha256":"3edf829c9d82d7518df23893e02c2440c522269bde03deada618dee09f4d9faa","source":{"kind":"arxiv","id":"2503.10872","version":2},"attestation_state":"computed","paper":{"title":"TAIJI: Textual Anchoring for Immunizing Jailbreak Images in Vision Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Jinwei Hu, Wenjie Ruan, Xiangyu Yin, Xiaowei Huang, Xingyu Zhao, Yi Dong, Yi Qi, Zhen Chen","submitted_at":"2025-03-13T20:39:31Z","abstract_excerpt":"Vision Language Models (VLMs) have demonstrated impressive inference capabilities, but remain vulnerable to jailbreak attacks that can induce harmful or unethical responses. Existing defence methods are predominantly white-box approaches that require access to model parameters and extensive modifications, making them costly and impractical for many real-world scenarios. Although some black-box defences have been proposed, they often impose input constraints or require multiple queries, limiting their effectiveness in safety-critical tasks such as autonomous driving. To address these challenges"},"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":"2503.10872","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-03-13T20:39:31Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"13bf3000b9407715814d2e23aadec6180414348db1108eae37e2f5f4dacf41cc","abstract_canon_sha256":"9e92d294b8c3c1f1516bb0fb669151e8d39bcf60c9a14b590dd31723386c4362"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:37:07.689019Z","signature_b64":"oZEHsSDrOoMgpH6wUN7P08TEKtV5cIUV061RWbcFPhrnU8mCZ97mWwTOzA6wXnSQS4ZqE3KfRsouOAdtwIQsDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3edf829c9d82d7518df23893e02c2440c522269bde03deada618dee09f4d9faa","last_reissued_at":"2026-07-05T10:37:07.688526Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:37:07.688526Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TAIJI: Textual Anchoring for Immunizing Jailbreak Images in Vision Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Jinwei Hu, Wenjie Ruan, Xiangyu Yin, Xiaowei Huang, Xingyu Zhao, Yi Dong, Yi Qi, Zhen Chen","submitted_at":"2025-03-13T20:39:31Z","abstract_excerpt":"Vision Language Models (VLMs) have demonstrated impressive inference capabilities, but remain vulnerable to jailbreak attacks that can induce harmful or unethical responses. Existing defence methods are predominantly white-box approaches that require access to model parameters and extensive modifications, making them costly and impractical for many real-world scenarios. Although some black-box defences have been proposed, they often impose input constraints or require multiple queries, limiting their effectiveness in safety-critical tasks such as autonomous driving. To address these challenges"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.10872","kind":"arxiv","version":2},"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/2503.10872/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":"2503.10872","created_at":"2026-07-05T10:37:07.688585+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.10872v2","created_at":"2026-07-05T10:37:07.688585+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.10872","created_at":"2026-07-05T10:37:07.688585+00:00"},{"alias_kind":"pith_short_12","alias_value":"H3PYFHE5QLLV","created_at":"2026-07-05T10:37:07.688585+00:00"},{"alias_kind":"pith_short_16","alias_value":"H3PYFHE5QLLVDDPS","created_at":"2026-07-05T10:37:07.688585+00:00"},{"alias_kind":"pith_short_8","alias_value":"H3PYFHE5","created_at":"2026-07-05T10:37:07.688585+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.04105","citing_title":"Enhancing Robustness of LLM-Driven Multi-Agent Systems through Randomized Smoothing","ref_index":29,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID","json":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID.json","graph_json":"https://pith.science/api/pith-number/H3PYFHE5QLLVDDPSHCJ6ALBEID/graph.json","events_json":"https://pith.science/api/pith-number/H3PYFHE5QLLVDDPSHCJ6ALBEID/events.json","paper":"https://pith.science/paper/H3PYFHE5"},"agent_actions":{"view_html":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID","download_json":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID.json","view_paper":"https://pith.science/paper/H3PYFHE5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.10872&json=true","fetch_graph":"https://pith.science/api/pith-number/H3PYFHE5QLLVDDPSHCJ6ALBEID/graph.json","fetch_events":"https://pith.science/api/pith-number/H3PYFHE5QLLVDDPSHCJ6ALBEID/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID/action/timestamp_anchor","attest_storage":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID/action/storage_attestation","attest_author":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID/action/author_attestation","sign_citation":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID/action/citation_signature","submit_replication":"https://pith.science/pith/H3PYFHE5QLLVDDPSHCJ6ALBEID/action/replication_record"}},"created_at":"2026-07-05T10:37:07.688585+00:00","updated_at":"2026-07-05T10:37:07.688585+00:00"}