{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FNJS74UCJGPG6SQRKLPIFDLJXZ","short_pith_number":"pith:FNJS74UC","schema_version":"1.0","canonical_sha256":"2b532ff282499e6f4a1152de828d69be6fb531273747602d5063962b9cffa568","source":{"kind":"arxiv","id":"2408.10848","version":4},"attestation_state":"computed","paper":{"title":"Perception-guided Jailbreak against Text-to-Image Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Geguang Pu, Le Liang, Run Wang, Tianlin Li, Weikai Miao, Xiaojun Jia, Yang Liu, Yihao Huang","submitted_at":"2024-08-20T13:40:25Z","abstract_excerpt":"In recent years, Text-to-Image (T2I) models have garnered significant attention due to their remarkable advancements. However, security concerns have emerged due to their potential to generate inappropriate or Not-Safe-For-Work (NSFW) images. In this paper, inspired by the observation that texts with different semantics can lead to similar human perceptions, we propose an LLM-driven perception-guided jailbreak method, termed PGJ. It is a black-box jailbreak method that requires no specific T2I model (model-free) and generates highly natural attack prompts. Specifically, we propose identifying "},"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":"2408.10848","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-08-20T13:40:25Z","cross_cats_sorted":[],"title_canon_sha256":"b092bad66dae9de00e0f2d0caeb59bcf27fc0b59ea19e08c25a13608aee1cff1","abstract_canon_sha256":"ca2ca8e83888403dc3ced9a348e08602e40e7ed7736b9e931c7d5588e532ade9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:11:23.297778Z","signature_b64":"OnDgPJVAaHf/AV18z4sRfLEwUkeSuJPLsPCAUKYtEpi3BVKFCk6Unby69cwHCNyz/rhungmWiopvjrlFNYycBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2b532ff282499e6f4a1152de828d69be6fb531273747602d5063962b9cffa568","last_reissued_at":"2026-07-05T10:11:23.297001Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:11:23.297001Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Perception-guided Jailbreak against Text-to-Image Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Geguang Pu, Le Liang, Run Wang, Tianlin Li, Weikai Miao, Xiaojun Jia, Yang Liu, Yihao Huang","submitted_at":"2024-08-20T13:40:25Z","abstract_excerpt":"In recent years, Text-to-Image (T2I) models have garnered significant attention due to their remarkable advancements. However, security concerns have emerged due to their potential to generate inappropriate or Not-Safe-For-Work (NSFW) images. In this paper, inspired by the observation that texts with different semantics can lead to similar human perceptions, we propose an LLM-driven perception-guided jailbreak method, termed PGJ. It is a black-box jailbreak method that requires no specific T2I model (model-free) and generates highly natural attack prompts. Specifically, we propose identifying "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.10848","kind":"arxiv","version":4},"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/2408.10848/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":"2408.10848","created_at":"2026-07-05T10:11:23.297102+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.10848v4","created_at":"2026-07-05T10:11:23.297102+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.10848","created_at":"2026-07-05T10:11:23.297102+00:00"},{"alias_kind":"pith_short_12","alias_value":"FNJS74UCJGPG","created_at":"2026-07-05T10:11:23.297102+00:00"},{"alias_kind":"pith_short_16","alias_value":"FNJS74UCJGPG6SQR","created_at":"2026-07-05T10:11:23.297102+00:00"},{"alias_kind":"pith_short_8","alias_value":"FNJS74UC","created_at":"2026-07-05T10:11:23.297102+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/FNJS74UCJGPG6SQRKLPIFDLJXZ","json":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ.json","graph_json":"https://pith.science/api/pith-number/FNJS74UCJGPG6SQRKLPIFDLJXZ/graph.json","events_json":"https://pith.science/api/pith-number/FNJS74UCJGPG6SQRKLPIFDLJXZ/events.json","paper":"https://pith.science/paper/FNJS74UC"},"agent_actions":{"view_html":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ","download_json":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ.json","view_paper":"https://pith.science/paper/FNJS74UC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.10848&json=true","fetch_graph":"https://pith.science/api/pith-number/FNJS74UCJGPG6SQRKLPIFDLJXZ/graph.json","fetch_events":"https://pith.science/api/pith-number/FNJS74UCJGPG6SQRKLPIFDLJXZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ/action/storage_attestation","attest_author":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ/action/author_attestation","sign_citation":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ/action/citation_signature","submit_replication":"https://pith.science/pith/FNJS74UCJGPG6SQRKLPIFDLJXZ/action/replication_record"}},"created_at":"2026-07-05T10:11:23.297102+00:00","updated_at":"2026-07-05T10:11:23.297102+00:00"}