{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HKGF2EBJAA3OCX2NB2ITG43T62","short_pith_number":"pith:HKGF2EBJ","schema_version":"1.0","canonical_sha256":"3a8c5d10290036e15f4d0e91337373f6bf5cab5382b2c8c558bf55dc07868ad8","source":{"kind":"arxiv","id":"2404.16873","version":2},"attestation_state":"computed","paper":{"title":"AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Anselm Paulus, Arman Zharmagambetov, Brandon Amos, Chuan Guo, Yuandong Tian","submitted_at":"2024-04-21T22:18:13Z","abstract_excerpt":"Large Language Models (LLMs) are vulnerable to jailbreaking attacks that lead to generation of inappropriate or harmful content. Manual red-teaming requires a time-consuming search for adversarial prompts, whereas automatic adversarial prompt generation often leads to semantically meaningless attacks that do not scale well. In this paper, we present a novel method that uses another LLM, called AdvPrompter, to generate human-readable adversarial prompts in seconds. AdvPrompter, which is trained using an alternating optimization algorithm, generates suffixes that veil the input instruction witho"},"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":"2404.16873","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-04-21T22:18:13Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"eb84caf788e8759590179a18061b3d31f8915610100ff4da935a05f06edc9f78","abstract_canon_sha256":"304e9d3b5e1fcc4474f08e3032ab4ab96f981316108ddbd732c1f5692fb93824"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:14:23.463067Z","signature_b64":"sQB5/A9O78wrPOpTY8p2XwAuDLxkDviPtlcOVWJeUm4pCdDwpF2GqWR3D/Da03Ueaz3oud1mK8IXL8psR9wSAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a8c5d10290036e15f4d0e91337373f6bf5cab5382b2c8c558bf55dc07868ad8","last_reissued_at":"2026-07-05T11:14:23.462625Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:14:23.462625Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CR","authors_text":"Anselm Paulus, Arman Zharmagambetov, Brandon Amos, Chuan Guo, Yuandong Tian","submitted_at":"2024-04-21T22:18:13Z","abstract_excerpt":"Large Language Models (LLMs) are vulnerable to jailbreaking attacks that lead to generation of inappropriate or harmful content. Manual red-teaming requires a time-consuming search for adversarial prompts, whereas automatic adversarial prompt generation often leads to semantically meaningless attacks that do not scale well. In this paper, we present a novel method that uses another LLM, called AdvPrompter, to generate human-readable adversarial prompts in seconds. AdvPrompter, which is trained using an alternating optimization algorithm, generates suffixes that veil the input instruction witho"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.16873","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/2404.16873/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":"2404.16873","created_at":"2026-07-05T11:14:23.462691+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.16873v2","created_at":"2026-07-05T11:14:23.462691+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.16873","created_at":"2026-07-05T11:14:23.462691+00:00"},{"alias_kind":"pith_short_12","alias_value":"HKGF2EBJAA3O","created_at":"2026-07-05T11:14:23.462691+00:00"},{"alias_kind":"pith_short_16","alias_value":"HKGF2EBJAA3OCX2N","created_at":"2026-07-05T11:14:23.462691+00:00"},{"alias_kind":"pith_short_8","alias_value":"HKGF2EBJ","created_at":"2026-07-05T11:14:23.462691+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":16,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.07918","citing_title":"Efficient Safety Alignment of Language Models via Latent Personality Traits","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2607.08180","citing_title":"Out of Sight: Compression-Aware Content Protection against Agentic Crawlers","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2606.25487","citing_title":"How Reliable Is Your Jailbreak Judge? Calibration and Adversarial Robustness of Automated ASR Scoring","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25487","citing_title":"How Reliable Is Your Jailbreak Judge? Calibration and Adversarial Robustness of Automated ASR Scoring","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17872","citing_title":"AnchorKV: Safety-Aware KV Cache Compression via Soft Penalty with a Refusal Anchor","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03647","citing_title":"Black-box, Adaptive, Efficient, Transferable, Harmful, Applicable... Attacks Are All You Need to Break LLMs","ref_index":34,"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":52,"is_internal_anchor":false},{"citing_arxiv_id":"2405.13068","citing_title":"Uncovering Logit Suppression Vulnerabilities in LLM Safety Alignment","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2510.07239","citing_title":"Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21362","citing_title":"LASH: Adaptive Semantic Hybridization for Black-Box Jailbreaking of Large Language Models","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17128","citing_title":"New Wide-Net-Casting Jailbreak Attacks Risk Large Models","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2407.04295","citing_title":"Jailbreak Attacks and Defenses Against Large Language Models: A Survey","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2603.24935","citing_title":"SABER: A Stealthy Agentic Black-Box Attack Framework for Vision-Language-Action Models","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18660","citing_title":"Evaluating Answer Leakage Robustness of LLM Tutors against Adversarial Student Attacks","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12548","citing_title":"DeepSeek Robustness Against Semantic-Character Dual-Space Mutated Prompt Injection","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12232","citing_title":"TEMPLATEFUZZ: Fine-Grained Chat Template Fuzzing for Jailbreaking and Red Teaming LLMs","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62","json":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62.json","graph_json":"https://pith.science/api/pith-number/HKGF2EBJAA3OCX2NB2ITG43T62/graph.json","events_json":"https://pith.science/api/pith-number/HKGF2EBJAA3OCX2NB2ITG43T62/events.json","paper":"https://pith.science/paper/HKGF2EBJ"},"agent_actions":{"view_html":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62","download_json":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62.json","view_paper":"https://pith.science/paper/HKGF2EBJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.16873&json=true","fetch_graph":"https://pith.science/api/pith-number/HKGF2EBJAA3OCX2NB2ITG43T62/graph.json","fetch_events":"https://pith.science/api/pith-number/HKGF2EBJAA3OCX2NB2ITG43T62/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62/action/storage_attestation","attest_author":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62/action/author_attestation","sign_citation":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62/action/citation_signature","submit_replication":"https://pith.science/pith/HKGF2EBJAA3OCX2NB2ITG43T62/action/replication_record"}},"created_at":"2026-07-05T11:14:23.462691+00:00","updated_at":"2026-07-05T11:14:23.462691+00:00"}