{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6GOACN6SM2HT36LRJVOJT4VRQO","short_pith_number":"pith:6GOACN6S","schema_version":"1.0","canonical_sha256":"f19c0137d2668f3df9714d5c99f2b183af0c2a73444d7a8cb7c3433255d37eb1","source":{"kind":"arxiv","id":"2510.07239","version":2},"attestation_state":"computed","paper":{"title":"Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alessandra Russo, Christos Ziakas, Nicholas Loo, Nishita Jain","submitted_at":"2025-10-08T17:06:20Z","abstract_excerpt":"Automated red-teaming has emerged as a scalable approach for auditing Large Language Models (LLMs) prior to deployment, yet existing approaches lack mechanisms to efficiently adapt to model-specific vulnerabilities at inference. We introduce Red-Bandit, a red-teaming framework that adapts online to identify and exploit model failure modes under distinct attack styles (e.g., manipulation, slang). Red-Bandit post-trains a set of parameter-efficient LoRA experts, each specialized for a particular attack style, using reinforcement learning that rewards the generation of unsafe prompts via a rule-b"},"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":"2510.07239","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-10-08T17:06:20Z","cross_cats_sorted":[],"title_canon_sha256":"6b9e19e114ecbce208127b7b80fae258ccd279a322fd395dcf705593e12d923e","abstract_canon_sha256":"044778dc06f7d271e51bc0c6845b02800d77c6f2d54eabb2e5b8f9350ace71ca"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:14.300167Z","signature_b64":"ULTUyMKQZPt3afdKkP6tfOveE7bxqTzjF/cm2fJ1apXkqE1koTBnQJwHRZgTXKEDkpnV8VHIDXqr80+Mha0pBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f19c0137d2668f3df9714d5c99f2b183af0c2a73444d7a8cb7c3433255d37eb1","last_reissued_at":"2026-05-20T00:04:14.299500Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:14.299500Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alessandra Russo, Christos Ziakas, Nicholas Loo, Nishita Jain","submitted_at":"2025-10-08T17:06:20Z","abstract_excerpt":"Automated red-teaming has emerged as a scalable approach for auditing Large Language Models (LLMs) prior to deployment, yet existing approaches lack mechanisms to efficiently adapt to model-specific vulnerabilities at inference. We introduce Red-Bandit, a red-teaming framework that adapts online to identify and exploit model failure modes under distinct attack styles (e.g., manipulation, slang). Red-Bandit post-trains a set of parameter-efficient LoRA experts, each specialized for a particular attack style, using reinforcement learning that rewards the generation of unsafe prompts via a rule-b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.07239","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/2510.07239/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":"2510.07239","created_at":"2026-05-20T00:04:14.299603+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.07239v2","created_at":"2026-05-20T00:04:14.299603+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.07239","created_at":"2026-05-20T00:04:14.299603+00:00"},{"alias_kind":"pith_short_12","alias_value":"6GOACN6SM2HT","created_at":"2026-05-20T00:04:14.299603+00:00"},{"alias_kind":"pith_short_16","alias_value":"6GOACN6SM2HT36LR","created_at":"2026-05-20T00:04:14.299603+00:00"},{"alias_kind":"pith_short_8","alias_value":"6GOACN6S","created_at":"2026-05-20T00:04:14.299603+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.08427","citing_title":"The Attacker in the Mirror: Breaking Self-Consistency in Safety via Anchored Bipolicy Self-Play","ref_index":41,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO","json":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO.json","graph_json":"https://pith.science/api/pith-number/6GOACN6SM2HT36LRJVOJT4VRQO/graph.json","events_json":"https://pith.science/api/pith-number/6GOACN6SM2HT36LRJVOJT4VRQO/events.json","paper":"https://pith.science/paper/6GOACN6S"},"agent_actions":{"view_html":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO","download_json":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO.json","view_paper":"https://pith.science/paper/6GOACN6S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.07239&json=true","fetch_graph":"https://pith.science/api/pith-number/6GOACN6SM2HT36LRJVOJT4VRQO/graph.json","fetch_events":"https://pith.science/api/pith-number/6GOACN6SM2HT36LRJVOJT4VRQO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO/action/storage_attestation","attest_author":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO/action/author_attestation","sign_citation":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO/action/citation_signature","submit_replication":"https://pith.science/pith/6GOACN6SM2HT36LRJVOJT4VRQO/action/replication_record"}},"created_at":"2026-05-20T00:04:14.299603+00:00","updated_at":"2026-05-20T00:04:14.299603+00:00"}