{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:3OV3ESHUDJ4ULZ34EXOM7VZWRV","short_pith_number":"pith:3OV3ESHU","schema_version":"1.0","canonical_sha256":"dbabb248f41a7945e77c25dccfd7368d7cfd03eb25bc81cd066e4f72dc12bc52","source":{"kind":"arxiv","id":"2401.17263","version":5},"attestation_state":"computed","paper":{"title":"Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attacks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Andy Zhou, Bo Li, Haohan Wang","submitted_at":"2024-01-30T18:56:08Z","abstract_excerpt":"Despite advances in AI alignment, large language models (LLMs) remain vulnerable to adversarial attacks or jailbreaking, in which adversaries can modify prompts to induce unwanted behavior. While some defenses have been proposed, they have not been adapted to newly proposed attacks and more challenging threat models. To address this, we propose an optimization-based objective for defending LLMs against jailbreaking attacks and an algorithm, Robust Prompt Optimization (RPO) to create robust system-level defenses. Our approach directly incorporates the adversary into the defensive objective and "},"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":"2401.17263","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-01-30T18:56:08Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CV"],"title_canon_sha256":"af54f738b7d646b9e17909e297213be56335b007132091a802c581242091af41","abstract_canon_sha256":"5c07699b0c5a295b978b7491959f8dab6bba40e63a903b793d85756919bf7adc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:32:41.854597Z","signature_b64":"HdObim88BOY7J27kZb0pk9JNGxqOhvmp1HCKPg5amtA66DunGUGJB1iFxy4XcF24Ltj6dV/tFcUM4Ja5eAqlAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dbabb248f41a7945e77c25dccfd7368d7cfd03eb25bc81cd066e4f72dc12bc52","last_reissued_at":"2026-07-05T09:32:41.854090Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:32:41.854090Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attacks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Andy Zhou, Bo Li, Haohan Wang","submitted_at":"2024-01-30T18:56:08Z","abstract_excerpt":"Despite advances in AI alignment, large language models (LLMs) remain vulnerable to adversarial attacks or jailbreaking, in which adversaries can modify prompts to induce unwanted behavior. While some defenses have been proposed, they have not been adapted to newly proposed attacks and more challenging threat models. To address this, we propose an optimization-based objective for defending LLMs against jailbreaking attacks and an algorithm, Robust Prompt Optimization (RPO) to create robust system-level defenses. Our approach directly incorporates the adversary into the defensive objective and "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.17263","kind":"arxiv","version":5},"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/2401.17263/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":"2401.17263","created_at":"2026-07-05T09:32:41.854150+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.17263v5","created_at":"2026-07-05T09:32:41.854150+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.17263","created_at":"2026-07-05T09:32:41.854150+00:00"},{"alias_kind":"pith_short_12","alias_value":"3OV3ESHUDJ4U","created_at":"2026-07-05T09:32:41.854150+00:00"},{"alias_kind":"pith_short_16","alias_value":"3OV3ESHUDJ4ULZ34","created_at":"2026-07-05T09:32:41.854150+00:00"},{"alias_kind":"pith_short_8","alias_value":"3OV3ESHU","created_at":"2026-07-05T09:32:41.854150+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.30783","citing_title":"Security--Fidelity Tradeoffs: The Hidden Cost of Prompt Injection Defense","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31748","citing_title":"Addressing Over-Refusal in LLMs with Competing Rewards","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2403.10559","citing_title":"Generative Models and Connected and Automated Vehicles: A Survey in Exploring the Intersection of Transportation and AI","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2511.12710","citing_title":"Evolve the Method, Not the Prompts: Evolutionary Synthesis of Jailbreak Attacks on LLMs","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2508.04204","citing_title":"ReasoningGuard: Safeguarding Large Reasoning Models with Inference-time Safety Aha Moments","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2407.04295","citing_title":"Jailbreak Attacks and Defenses Against Large Language Models: A Survey","ref_index":121,"is_internal_anchor":false},{"citing_arxiv_id":"2402.17177","citing_title":"Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Models","ref_index":115,"is_internal_anchor":false},{"citing_arxiv_id":"2310.08419","citing_title":"Jailbreaking Black Box Large Language Models in Twenty Queries","ref_index":55,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV","json":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV.json","graph_json":"https://pith.science/api/pith-number/3OV3ESHUDJ4ULZ34EXOM7VZWRV/graph.json","events_json":"https://pith.science/api/pith-number/3OV3ESHUDJ4ULZ34EXOM7VZWRV/events.json","paper":"https://pith.science/paper/3OV3ESHU"},"agent_actions":{"view_html":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV","download_json":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV.json","view_paper":"https://pith.science/paper/3OV3ESHU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.17263&json=true","fetch_graph":"https://pith.science/api/pith-number/3OV3ESHUDJ4ULZ34EXOM7VZWRV/graph.json","fetch_events":"https://pith.science/api/pith-number/3OV3ESHUDJ4ULZ34EXOM7VZWRV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV/action/storage_attestation","attest_author":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV/action/author_attestation","sign_citation":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV/action/citation_signature","submit_replication":"https://pith.science/pith/3OV3ESHUDJ4ULZ34EXOM7VZWRV/action/replication_record"}},"created_at":"2026-07-05T09:32:41.854150+00:00","updated_at":"2026-07-05T09:32:41.854150+00:00"}