{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:XAH345MIO3UQYALFZEENVWSHJO","short_pith_number":"pith:XAH345MI","schema_version":"1.0","canonical_sha256":"b80fbe758876e90c0165c908dada474b9826daaad30659ef5f7da5985e407be3","source":{"kind":"arxiv","id":"2406.01637","version":2},"attestation_state":"computed","paper":{"title":"Teams of LLM Agents can Exploit Zero-Day Vulnerabilities","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.MA","authors_text":"Akul Gupta, Antony Kellermann, Daniel Kang, Philip Li, Richard Fang, Rohan Bindu, Yuxuan Zhu","submitted_at":"2024-06-02T16:25:26Z","abstract_excerpt":"LLM agents have become increasingly sophisticated, especially in the realm of cybersecurity. Researchers have shown that LLM agents can exploit real-world vulnerabilities when given a description of the vulnerability and toy capture-the-flag problems. However, these agents still perform poorly on real-world vulnerabilities that are unknown to the agent ahead of time (zero-day vulnerabilities).\n  In this work, we show that teams of LLM agents can exploit real-world, zero-day vulnerabilities. Prior agents struggle with exploring many different vulnerabilities and long-range planning when used al"},"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":"2406.01637","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.MA","submitted_at":"2024-06-02T16:25:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"a766dfd877a32de7f27e4ca889b244a0ff4cb2b2d998ebb95d6d26556e773b43","abstract_canon_sha256":"c9208c3a4974418675acc35858151a583cab81366a0165245460f1e8b51949d6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:41:12.860638Z","signature_b64":"vQ0Q/3smvQqcl6GcASKmiwKwLl9R5UADdukarpDzmgjK0vxiRRHOiMC8su8RcpFEWa288S2cGjzPyK9C+h8TDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b80fbe758876e90c0165c908dada474b9826daaad30659ef5f7da5985e407be3","last_reissued_at":"2026-07-05T10:41:12.860150Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:41:12.860150Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Teams of LLM Agents can Exploit Zero-Day Vulnerabilities","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.MA","authors_text":"Akul Gupta, Antony Kellermann, Daniel Kang, Philip Li, Richard Fang, Rohan Bindu, Yuxuan Zhu","submitted_at":"2024-06-02T16:25:26Z","abstract_excerpt":"LLM agents have become increasingly sophisticated, especially in the realm of cybersecurity. Researchers have shown that LLM agents can exploit real-world vulnerabilities when given a description of the vulnerability and toy capture-the-flag problems. However, these agents still perform poorly on real-world vulnerabilities that are unknown to the agent ahead of time (zero-day vulnerabilities).\n  In this work, we show that teams of LLM agents can exploit real-world, zero-day vulnerabilities. Prior agents struggle with exploring many different vulnerabilities and long-range planning when used al"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01637","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/2406.01637/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":"2406.01637","created_at":"2026-07-05T10:41:12.860207+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.01637v2","created_at":"2026-07-05T10:41:12.860207+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01637","created_at":"2026-07-05T10:41:12.860207+00:00"},{"alias_kind":"pith_short_12","alias_value":"XAH345MIO3UQ","created_at":"2026-07-05T10:41:12.860207+00:00"},{"alias_kind":"pith_short_16","alias_value":"XAH345MIO3UQYALF","created_at":"2026-07-05T10:41:12.860207+00:00"},{"alias_kind":"pith_short_8","alias_value":"XAH345MI","created_at":"2026-07-05T10:41:12.860207+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":20,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07109","citing_title":"Certifying Ghosts: How Cybersecurity AI Agents Break the EU Cyber Resilience Act","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2606.05567","citing_title":"ZERO-APT: A Closed-Loop Adversarial Framework for LLM-Driven Automated Penetration Testing under Intelligent Defense","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03453","citing_title":"FORGE: Multi-Agent Graduated Exploitation and Detection Engineering","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00341","citing_title":"ROGUE: Misaligned Agent Behavior Arising from Ordinary Computer Use","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29981","citing_title":"Hephaestus: Toward a Cybersecurity AI Scientist","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30096","citing_title":"How Reliable Are AI Attackers Against a Fixed Vulnerable Target? A 400-Run Empirical Study of LLM Penetration Testing Consistency","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17416","citing_title":"Benchmarking Mythos-Linked Bug Rediscovery","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2509.06921","citing_title":"Neuro-Symbolic AI for Cybersecurity: State of the Art, Challenges, and Opportunities","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2509.13021","citing_title":"xOffense: An Autonomous Multi-Agent Framework for Penetration Testing with Domain-Adapted Large Language Models","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2512.16280","citing_title":"Love, Lies, and Language Models: Investigating AI's Role in Romance-Baiting Scams","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2512.22753","citing_title":"From Rookie to Expert: Manipulating LLMs for Automated Vulnerability Exploitation in Enterprise Software","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24184","citing_title":"Dynamic Cyber Ranges","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23338","citing_title":"A Systematic Survey of Security Threats and Defenses in LLM-Based AI Agents: A Layered Attack Surface Framework","ref_index":99,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06601","citing_title":"Patch2Vuln: Agentic Reconstruction of Vulnerabilities from Linux Distribution Binary Patches","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05531","citing_title":"Beyond Collection: Measuring the Detection Efficacy of Modern Security Logging Standards","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01186","citing_title":"Trace: Unmasking AI Attack Agents Through Terminal Behavior Fingerprinting","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18718","citing_title":"Towards Optimal Agentic Architectures for Offensive Security Tasks","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06713","citing_title":"Agentic AI and the Industrialization of Cyber Offense: Forecast, Consequences, and Defensive Priorities for Enterprises and the Mittelstand","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05130","citing_title":"A Multi-Agent Framework for Automated Exploit Generation with Constraint-Guided Comprehension and Reflection","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20179","citing_title":"Taint-Style Vulnerability Detection and Confirmation for Node.js Packages Using LLM Agent Reasoning","ref_index":58,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO","json":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO.json","graph_json":"https://pith.science/api/pith-number/XAH345MIO3UQYALFZEENVWSHJO/graph.json","events_json":"https://pith.science/api/pith-number/XAH345MIO3UQYALFZEENVWSHJO/events.json","paper":"https://pith.science/paper/XAH345MI"},"agent_actions":{"view_html":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO","download_json":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO.json","view_paper":"https://pith.science/paper/XAH345MI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.01637&json=true","fetch_graph":"https://pith.science/api/pith-number/XAH345MIO3UQYALFZEENVWSHJO/graph.json","fetch_events":"https://pith.science/api/pith-number/XAH345MIO3UQYALFZEENVWSHJO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO/action/storage_attestation","attest_author":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO/action/author_attestation","sign_citation":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO/action/citation_signature","submit_replication":"https://pith.science/pith/XAH345MIO3UQYALFZEENVWSHJO/action/replication_record"}},"created_at":"2026-07-05T10:41:12.860207+00:00","updated_at":"2026-07-05T10:41:12.860207+00:00"}