{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:R7DSUVG7I46SN3UDVMI3XW2HCQ","short_pith_number":"pith:R7DSUVG7","schema_version":"1.0","canonical_sha256":"8fc72a54df473d26ee83ab11bbdb4714071db9d17273e090c17cc9d26e9485a3","source":{"kind":"arxiv","id":"2406.04031","version":2},"attestation_state":"computed","paper":{"title":"Jailbreak Vision Language Models via Bi-Modal Adversarial Prompt","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CV","authors_text":"Aishan Liu, Dacheng Tao, Siyuan Liang, Tianyuan Zhang, Xianglong Liu, Zhengmin Yu, Zonghao Ying","submitted_at":"2024-06-06T13:00:42Z","abstract_excerpt":"In the realm of large vision language models (LVLMs), jailbreak attacks serve as a red-teaming approach to bypass guardrails and uncover safety implications. Existing jailbreaks predominantly focus on the visual modality, perturbing solely visual inputs in the prompt for attacks. However, they fall short when confronted with aligned models that fuse visual and textual features simultaneously for generation. To address this limitation, this paper introduces the Bi-Modal Adversarial Prompt Attack (BAP), which executes jailbreaks by optimizing textual and visual prompts cohesively. Initially, we "},"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":"2406.04031","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-06-06T13:00:42Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"c37f5e7d256542e9265e7e09f733c47fc149d9beb992fb1c8058cc7fc2a99d17","abstract_canon_sha256":"deee94bb00921f4176d95d4c3054f63af108afdb24c523680e65dcebb226d488"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:38:38.457016Z","signature_b64":"3lUzSg0twbALfYwyRnyYBqGkUm9FYEuq7sqzqXAMZ4v0+jD7XOfFch47NcvPk0URc0zgjEAnEZA2nWfz0vn7Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8fc72a54df473d26ee83ab11bbdb4714071db9d17273e090c17cc9d26e9485a3","last_reissued_at":"2026-07-05T08:38:38.456606Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:38:38.456606Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Jailbreak Vision Language Models via Bi-Modal Adversarial Prompt","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CV","authors_text":"Aishan Liu, Dacheng Tao, Siyuan Liang, Tianyuan Zhang, Xianglong Liu, Zhengmin Yu, Zonghao Ying","submitted_at":"2024-06-06T13:00:42Z","abstract_excerpt":"In the realm of large vision language models (LVLMs), jailbreak attacks serve as a red-teaming approach to bypass guardrails and uncover safety implications. Existing jailbreaks predominantly focus on the visual modality, perturbing solely visual inputs in the prompt for attacks. However, they fall short when confronted with aligned models that fuse visual and textual features simultaneously for generation. To address this limitation, this paper introduces the Bi-Modal Adversarial Prompt Attack (BAP), which executes jailbreaks by optimizing textual and visual prompts cohesively. Initially, we "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.04031","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/2406.04031/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":"2406.04031","created_at":"2026-07-05T08:38:38.456666+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.04031v2","created_at":"2026-07-05T08:38:38.456666+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.04031","created_at":"2026-07-05T08:38:38.456666+00:00"},{"alias_kind":"pith_short_12","alias_value":"R7DSUVG7I46S","created_at":"2026-07-05T08:38:38.456666+00:00"},{"alias_kind":"pith_short_16","alias_value":"R7DSUVG7I46SN3UD","created_at":"2026-07-05T08:38:38.456666+00:00"},{"alias_kind":"pith_short_8","alias_value":"R7DSUVG7","created_at":"2026-07-05T08:38:38.456666+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25194","citing_title":"Localization then Neutralization: Gradient-guided Token Suppression against Visual Prompt Injection Attack","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26566","citing_title":"Adversarial Diffusion Across Modalities: A Fusion Survey of Attacks, Defenses, and Evaluation for Text, Vision, and Vision-Language Models","ref_index":77,"is_internal_anchor":false},{"citing_arxiv_id":"2411.18275","citing_title":"Visual Adversarial Attack on Vision-Language Models for Autonomous Driving","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21540","citing_title":"PRISM: Programmatic Reasoning with Image Sequence Manipulation for LVLM Jailbreaking","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04261","citing_title":"Laundering AI Authority with Adversarial Examples","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04488","citing_title":"A Patch-based Cross-view Regularized Framework for Backdoor Defense in Multimodal Large Language Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04630","citing_title":"Multimodal Backdoor Attack on VLMs for Autonomous Driving via Graffiti and Cross-Lingual Triggers","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18803","citing_title":"LLM-as-Judge Framework for Evaluating Tone-Induced Hallucination in Vision-Language Models","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ","json":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ.json","graph_json":"https://pith.science/api/pith-number/R7DSUVG7I46SN3UDVMI3XW2HCQ/graph.json","events_json":"https://pith.science/api/pith-number/R7DSUVG7I46SN3UDVMI3XW2HCQ/events.json","paper":"https://pith.science/paper/R7DSUVG7"},"agent_actions":{"view_html":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ","download_json":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ.json","view_paper":"https://pith.science/paper/R7DSUVG7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.04031&json=true","fetch_graph":"https://pith.science/api/pith-number/R7DSUVG7I46SN3UDVMI3XW2HCQ/graph.json","fetch_events":"https://pith.science/api/pith-number/R7DSUVG7I46SN3UDVMI3XW2HCQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ/action/storage_attestation","attest_author":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ/action/author_attestation","sign_citation":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ/action/citation_signature","submit_replication":"https://pith.science/pith/R7DSUVG7I46SN3UDVMI3XW2HCQ/action/replication_record"}},"created_at":"2026-07-05T08:38:38.456666+00:00","updated_at":"2026-07-05T08:38:38.456666+00:00"}