{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:RAWEPYPDW7RCK77CPU7OQRP37G","short_pith_number":"pith:RAWEPYPD","schema_version":"1.0","canonical_sha256":"882c47e1e3b7e2257fe27d3ee845fbf9ad85cef855c2a9354eae58a244d2a883","source":{"kind":"arxiv","id":"2405.19103","version":1},"attestation_state":"computed","paper":{"title":"Voice Jailbreak Attacks Against GPT-4o","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Michael Backes, Xinyue Shen, Yang Zhang, Yixin Wu","submitted_at":"2024-05-29T14:07:44Z","abstract_excerpt":"Recently, the concept of artificial assistants has evolved from science fiction into real-world applications. GPT-4o, the newest multimodal large language model (MLLM) across audio, vision, and text, has further blurred the line between fiction and reality by enabling more natural human-computer interactions. However, the advent of GPT-4o's voice mode may also introduce a new attack surface. In this paper, we present the first systematic measurement of jailbreak attacks against the voice mode of GPT-4o. We show that GPT-4o demonstrates good resistance to forbidden questions and text jailbreak "},"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":"2405.19103","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-05-29T14:07:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"1c19dbe06c73091d58d83a05ddf07c2e4f9502256bb98ce347a3e653c60aaf02","abstract_canon_sha256":"d539941dc65e82db91b4db5b302d0d488df94743e065f78fbe739a61e6b757f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:24:47.913213Z","signature_b64":"ougFge+JzE2z0PwxSbDAQiOdtNN0Sy1aZw6HjCBE/o5DLp+I1xF8V9mBq5kEBV880fDK7n6Q5n0F3ej0MPyyDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"882c47e1e3b7e2257fe27d3ee845fbf9ad85cef855c2a9354eae58a244d2a883","last_reissued_at":"2026-07-05T08:24:47.912728Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:24:47.912728Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Voice Jailbreak Attacks Against GPT-4o","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Michael Backes, Xinyue Shen, Yang Zhang, Yixin Wu","submitted_at":"2024-05-29T14:07:44Z","abstract_excerpt":"Recently, the concept of artificial assistants has evolved from science fiction into real-world applications. GPT-4o, the newest multimodal large language model (MLLM) across audio, vision, and text, has further blurred the line between fiction and reality by enabling more natural human-computer interactions. However, the advent of GPT-4o's voice mode may also introduce a new attack surface. In this paper, we present the first systematic measurement of jailbreak attacks against the voice mode of GPT-4o. We show that GPT-4o demonstrates good resistance to forbidden questions and text jailbreak "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.19103","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/2405.19103/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":"2405.19103","created_at":"2026-07-05T08:24:47.912789+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.19103v1","created_at":"2026-07-05T08:24:47.912789+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.19103","created_at":"2026-07-05T08:24:47.912789+00:00"},{"alias_kind":"pith_short_12","alias_value":"RAWEPYPDW7RC","created_at":"2026-07-05T08:24:47.912789+00:00"},{"alias_kind":"pith_short_16","alias_value":"RAWEPYPDW7RCK77C","created_at":"2026-07-05T08:24:47.912789+00:00"},{"alias_kind":"pith_short_8","alias_value":"RAWEPYPD","created_at":"2026-07-05T08:24:47.912789+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06037","citing_title":"SpeechJBB: Probing Safety Alignment and Comprehension in Large Audio Language Models under Code-Switched Speech","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04483","citing_title":"Off-Distribution Voices: Fanfiction Subgenres as Universal Vernacular Jailbreaks for Aligned LLMs","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30031","citing_title":"Audio Jailbreaks in Large Audio-Language Models: Taxonomy, Attack-Defense Analysis, and Cost-Aware Evaluation","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20519","citing_title":"Codec-Robust Attacks on Audio LLMs","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2505.15957","citing_title":"Towards Holistic Evaluation of Large Audio-Language Models: A Comprehensive Survey","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20519","citing_title":"Codec-Robust Attacks on Audio LLMs","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18168","citing_title":"Acoustic Interference: A New Paradigm Weaponizing Acoustic Latent Semantic for Universal Jailbreak against Large Audio Language Models","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11309","citing_title":"The Salami Slicing Threat: Exploiting Cumulative Risks in LLM Systems","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14604","citing_title":"Hijacking Large Audio-Language Models via Context-Agnostic and Imperceptible Auditory Prompt Injection","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16659","citing_title":"Benign Fine-Tuning Breaks Safety Alignment in Audio LLMs","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G","json":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G.json","graph_json":"https://pith.science/api/pith-number/RAWEPYPDW7RCK77CPU7OQRP37G/graph.json","events_json":"https://pith.science/api/pith-number/RAWEPYPDW7RCK77CPU7OQRP37G/events.json","paper":"https://pith.science/paper/RAWEPYPD"},"agent_actions":{"view_html":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G","download_json":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G.json","view_paper":"https://pith.science/paper/RAWEPYPD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.19103&json=true","fetch_graph":"https://pith.science/api/pith-number/RAWEPYPDW7RCK77CPU7OQRP37G/graph.json","fetch_events":"https://pith.science/api/pith-number/RAWEPYPDW7RCK77CPU7OQRP37G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G/action/storage_attestation","attest_author":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G/action/author_attestation","sign_citation":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G/action/citation_signature","submit_replication":"https://pith.science/pith/RAWEPYPDW7RCK77CPU7OQRP37G/action/replication_record"}},"created_at":"2026-07-05T08:24:47.912789+00:00","updated_at":"2026-07-05T08:24:47.912789+00:00"}