{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:55TDSEUFROACICSEQKMVESBHB5","short_pith_number":"pith:55TDSEUF","schema_version":"1.0","canonical_sha256":"ef663912858b80240a4482995248270f579752b49f40b3d56d77076b84c464b3","source":{"kind":"arxiv","id":"2602.16346","version":3},"attestation_state":"computed","paper":{"title":"Helpful to a Fault: Measuring Illicit Assistance in Multi-Turn, Multilingual LLM Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Antoine Bosselut, Ayush K Tarun, Maksym Andriushchenko, Murari Mandal, Nivya Talokar","submitted_at":"2026-02-18T10:31:19Z","abstract_excerpt":"LLM-based agents execute real-world workflows via tools and memory. These affordances enable ill-intended adversaries to also use these agents to carry out complex misuse scenarios. Existing agent misuse benchmarks largely test single-prompt instructions, leaving a gap in measuring how agents end up helping with harmful or illegal tasks over multiple turns. We introduce STING (Sequential Testing of Illicit N-step Goal execution), an automated red-teaming framework that constructs a step-by-step illicit plan grounded in a benign persona and iteratively probes a target agent with adaptive follow"},"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":"2602.16346","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-02-18T10:31:19Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"428d7b8c0a0dedf95e4b92bbf3a76bf383c81b8f24fc1ffa3a1c254bd4cfb6e8","abstract_canon_sha256":"00625ead66de3940eb589c22ff3b62da5ffd374589b3a407ab8c3c408b726b5c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:09.793188Z","signature_b64":"cwLKfVB1WrI/auVxWfrvY3ZjbdpSaxkdVexafhXfnMPfWQnlLmxDnm+Xi39hrlZdQWm3xhBvpDmcwNX7qT5DDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ef663912858b80240a4482995248270f579752b49f40b3d56d77076b84c464b3","last_reissued_at":"2026-05-20T00:02:09.792419Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:09.792419Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Helpful to a Fault: Measuring Illicit Assistance in Multi-Turn, Multilingual LLM Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Antoine Bosselut, Ayush K Tarun, Maksym Andriushchenko, Murari Mandal, Nivya Talokar","submitted_at":"2026-02-18T10:31:19Z","abstract_excerpt":"LLM-based agents execute real-world workflows via tools and memory. These affordances enable ill-intended adversaries to also use these agents to carry out complex misuse scenarios. Existing agent misuse benchmarks largely test single-prompt instructions, leaving a gap in measuring how agents end up helping with harmful or illegal tasks over multiple turns. We introduce STING (Sequential Testing of Illicit N-step Goal execution), an automated red-teaming framework that constructs a step-by-step illicit plan grounded in a benign persona and iteratively probes a target agent with adaptive follow"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.16346","kind":"arxiv","version":3},"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/2602.16346/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":"2602.16346","created_at":"2026-05-20T00:02:09.792545+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.16346v3","created_at":"2026-05-20T00:02:09.792545+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.16346","created_at":"2026-05-20T00:02:09.792545+00:00"},{"alias_kind":"pith_short_12","alias_value":"55TDSEUFROAC","created_at":"2026-05-20T00:02:09.792545+00:00"},{"alias_kind":"pith_short_16","alias_value":"55TDSEUFROACICSE","created_at":"2026-05-20T00:02:09.792545+00:00"},{"alias_kind":"pith_short_8","alias_value":"55TDSEUF","created_at":"2026-05-20T00:02:09.792545+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5","json":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5.json","graph_json":"https://pith.science/api/pith-number/55TDSEUFROACICSEQKMVESBHB5/graph.json","events_json":"https://pith.science/api/pith-number/55TDSEUFROACICSEQKMVESBHB5/events.json","paper":"https://pith.science/paper/55TDSEUF"},"agent_actions":{"view_html":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5","download_json":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5.json","view_paper":"https://pith.science/paper/55TDSEUF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.16346&json=true","fetch_graph":"https://pith.science/api/pith-number/55TDSEUFROACICSEQKMVESBHB5/graph.json","fetch_events":"https://pith.science/api/pith-number/55TDSEUFROACICSEQKMVESBHB5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5/action/storage_attestation","attest_author":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5/action/author_attestation","sign_citation":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5/action/citation_signature","submit_replication":"https://pith.science/pith/55TDSEUFROACICSEQKMVESBHB5/action/replication_record"}},"created_at":"2026-05-20T00:02:09.792545+00:00","updated_at":"2026-05-20T00:02:09.792545+00:00"}