{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:WXUUJ37HREPD5K5J6XNR344F2M","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"9d484bd604426974e92160d2702f537a8f88a9e264757b841c3cb88ae04b3b4a","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-26T17:27:32Z","title_canon_sha256":"423542e7cd2c7d2cdccab676d1982b0828692349fda56cedb8a75dbfc52f96a7"},"schema_version":"1.0","source":{"id":"2505.21556","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.21556","created_at":"2026-07-05T11:10:59Z"},{"alias_kind":"arxiv_version","alias_value":"2505.21556v1","created_at":"2026-07-05T11:10:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.21556","created_at":"2026-07-05T11:10:59Z"},{"alias_kind":"pith_short_12","alias_value":"WXUUJ37HREPD","created_at":"2026-07-05T11:10:59Z"},{"alias_kind":"pith_short_16","alias_value":"WXUUJ37HREPD5K5J","created_at":"2026-07-05T11:10:59Z"},{"alias_kind":"pith_short_8","alias_value":"WXUUJ37H","created_at":"2026-07-05T11:10:59Z"}],"graph_snapshots":[{"event_id":"sha256:e95e09e0f6d7394fa47ab3469238e5b57d2cd85bc8fbfba75694aac559e76f00","target":"graph","created_at":"2026-07-05T11:10:59Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2505.21556/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Optimization-based jailbreaks typically adopt the Toxic-Continuation setting in large vision-language models (LVLMs), following the standard next-token prediction objective. In this setting, an adversarial image is optimized to make the model predict the next token of a toxic prompt. However, we find that the Toxic-Continuation paradigm is effective at continuing already-toxic inputs, but struggles to induce safety misalignment when explicit toxic signals are absent. We propose a new paradigm: Benign-to-Toxic (B2T) jailbreak. Unlike prior work, we optimize adversarial images to induce toxic ou","authors_text":"Changick Kim, Hee-Seon Kim, Kihyun Kim, Minbeom Kim, Wonjun Lee","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-26T17:27:32Z","title":"Benign-to-Toxic Jailbreaking: Inducing Harmful Responses from Harmless Prompts"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.21556","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:33b32df304b2e2620e02890df4cde82f0c36f11ebf404c798972f42ba889ea25","target":"record","created_at":"2026-07-05T11:10:59Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"9d484bd604426974e92160d2702f537a8f88a9e264757b841c3cb88ae04b3b4a","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-26T17:27:32Z","title_canon_sha256":"423542e7cd2c7d2cdccab676d1982b0828692349fda56cedb8a75dbfc52f96a7"},"schema_version":"1.0","source":{"id":"2505.21556","kind":"arxiv","version":1}},"canonical_sha256":"b5e944efe7891e3eaba9f5db1df385d301983519fe21724685afa79e0a1531c9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b5e944efe7891e3eaba9f5db1df385d301983519fe21724685afa79e0a1531c9","first_computed_at":"2026-07-05T11:10:59.851566Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:10:59.851566Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"k+EV1Atp/IHzG9+7Yo2OTTnMfYWtmz1LigBEiNGQxEx+/mL2EhCVmGm5Ne11VTr4tgODJccpT2O3oQZuJqZRBw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:10:59.852008Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.21556","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:33b32df304b2e2620e02890df4cde82f0c36f11ebf404c798972f42ba889ea25","sha256:e95e09e0f6d7394fa47ab3469238e5b57d2cd85bc8fbfba75694aac559e76f00"],"state_sha256":"67bc82cbd8f8f2c5b34285271eb319c8ddfa9085c9f28da229be08ed6fd21bd2"}