{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:VW7DQR7BRZMQPPZCZ6T5DDWI75","short_pith_number":"pith:VW7DQR7B","schema_version":"1.0","canonical_sha256":"adbe3847e18e5907bf22cfa7d18ec8ff45f45a255f787a8c91b5cfee4f745e32","source":{"kind":"arxiv","id":"2307.15008","version":1},"attestation_state":"computed","paper":{"title":"A LLM Assisted Exploitation of AI-Guardian","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CR","authors_text":"Nicholas Carlini","submitted_at":"2023-07-20T17:33:25Z","abstract_excerpt":"Large language models (LLMs) are now highly capable at a diverse range of tasks. This paper studies whether or not GPT-4, one such LLM, is capable of assisting researchers in the field of adversarial machine learning. As a case study, we evaluate the robustness of AI-Guardian, a recent defense to adversarial examples published at IEEE S&P 2023, a top computer security conference. We completely break this defense: the proposed scheme does not increase robustness compared to an undefended baseline.\n  We write none of the code to attack this model, and instead prompt GPT-4 to implement all attack"},"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":"2307.15008","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2023-07-20T17:33:25Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"1686dea1c71aa91a6bc22135fc9a6fc4c9dc9221dad26e82de6cf5a01c2e2cd6","abstract_canon_sha256":"d4180c1741998d34bbb09e3e534ff4783033d1f8156d830651e479f09cc5e053"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:35:20.713276Z","signature_b64":"8a0IFkXGg3aPCLCggV+d+86+dq+JgRrr2RCt6cJYcMVallp2e0rM2+zv3qrMY7Hf9IWGYT/IrjjjgMQgUTMXDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"adbe3847e18e5907bf22cfa7d18ec8ff45f45a255f787a8c91b5cfee4f745e32","last_reissued_at":"2026-07-05T06:35:20.712874Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:35:20.712874Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A LLM Assisted Exploitation of AI-Guardian","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CR","authors_text":"Nicholas Carlini","submitted_at":"2023-07-20T17:33:25Z","abstract_excerpt":"Large language models (LLMs) are now highly capable at a diverse range of tasks. This paper studies whether or not GPT-4, one such LLM, is capable of assisting researchers in the field of adversarial machine learning. As a case study, we evaluate the robustness of AI-Guardian, a recent defense to adversarial examples published at IEEE S&P 2023, a top computer security conference. We completely break this defense: the proposed scheme does not increase robustness compared to an undefended baseline.\n  We write none of the code to attack this model, and instead prompt GPT-4 to implement all attack"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2307.15008","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/2307.15008/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":"2307.15008","created_at":"2026-07-05T06:35:20.712930+00:00"},{"alias_kind":"arxiv_version","alias_value":"2307.15008v1","created_at":"2026-07-05T06:35:20.712930+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2307.15008","created_at":"2026-07-05T06:35:20.712930+00:00"},{"alias_kind":"pith_short_12","alias_value":"VW7DQR7BRZMQ","created_at":"2026-07-05T06:35:20.712930+00:00"},{"alias_kind":"pith_short_16","alias_value":"VW7DQR7BRZMQPPZC","created_at":"2026-07-05T06:35:20.712930+00:00"},{"alias_kind":"pith_short_8","alias_value":"VW7DQR7B","created_at":"2026-07-05T06:35:20.712930+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00012","citing_title":"PRA-RAG: Provably Robust Aggregation in Retrieval-Augmented Generation against Retrieval Corruption","ref_index":53,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75","json":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75.json","graph_json":"https://pith.science/api/pith-number/VW7DQR7BRZMQPPZCZ6T5DDWI75/graph.json","events_json":"https://pith.science/api/pith-number/VW7DQR7BRZMQPPZCZ6T5DDWI75/events.json","paper":"https://pith.science/paper/VW7DQR7B"},"agent_actions":{"view_html":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75","download_json":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75.json","view_paper":"https://pith.science/paper/VW7DQR7B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2307.15008&json=true","fetch_graph":"https://pith.science/api/pith-number/VW7DQR7BRZMQPPZCZ6T5DDWI75/graph.json","fetch_events":"https://pith.science/api/pith-number/VW7DQR7BRZMQPPZCZ6T5DDWI75/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75/action/storage_attestation","attest_author":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75/action/author_attestation","sign_citation":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75/action/citation_signature","submit_replication":"https://pith.science/pith/VW7DQR7BRZMQPPZCZ6T5DDWI75/action/replication_record"}},"created_at":"2026-07-05T06:35:20.712930+00:00","updated_at":"2026-07-05T06:35:20.712930+00:00"}