{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:EA3645LHD7CA46XOL7T3OSJLRJ","short_pith_number":"pith:EA3645LH","schema_version":"1.0","canonical_sha256":"2037ee75671fc40e7aee5fe7b7492b8a74b40d33adcd9440cb78518c77ec5672","source":{"kind":"arxiv","id":"2606.05523","version":1},"attestation_state":"computed","paper":{"title":"CHASE: Adversarial Red-Blue Teaming for Improving LLM Safety using Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Joshi, Alan Niu, Chris Yoo, Ishmanbir Singh, Rahul Markasserithodi, Yuekang Li","submitted_at":"2026-06-04T00:06:13Z","abstract_excerpt":"Despite advances in safety alignment, prompt-rewriting attacks such as persona modulation, fictional framing and persuasion-based reformulation, can bypass safety filters even on frontier models. Existing defenses either rely on non-scalable human curation or white-box optimisation that overfits to specific model internals, leaving aligned models brittle against the very class of adaptive black-box adversaries they will face in deployment. To address this gap, we introduce CHASE (Co-evolutionary Hardening through Adversarial Safety-Escalation), a closed-loop red-blue teaming framework in which"},"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":"2606.05523","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-04T00:06:13Z","cross_cats_sorted":[],"title_canon_sha256":"d3b81b860a044f5c0fc0ff26e63d7773854f0f144fdede18a3bf3a6c93421399","abstract_canon_sha256":"8704c38e257ee4ceaff773d5f2467f2d794a2c23852e85b48f51416cfccefd2c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:14:53.727677Z","signature_b64":"6EO1w3zn7lNQJmTNNWE3pckE9D3o55zYi5CwzBJf0CQXyL1jPT9tew3M/9Aey+KDRc81FWVSqKSG+7PfXU/gDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2037ee75671fc40e7aee5fe7b7492b8a74b40d33adcd9440cb78518c77ec5672","last_reissued_at":"2026-06-05T01:14:53.727220Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:14:53.727220Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CHASE: Adversarial Red-Blue Teaming for Improving LLM Safety using Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Joshi, Alan Niu, Chris Yoo, Ishmanbir Singh, Rahul Markasserithodi, Yuekang Li","submitted_at":"2026-06-04T00:06:13Z","abstract_excerpt":"Despite advances in safety alignment, prompt-rewriting attacks such as persona modulation, fictional framing and persuasion-based reformulation, can bypass safety filters even on frontier models. Existing defenses either rely on non-scalable human curation or white-box optimisation that overfits to specific model internals, leaving aligned models brittle against the very class of adaptive black-box adversaries they will face in deployment. To address this gap, we introduce CHASE (Co-evolutionary Hardening through Adversarial Safety-Escalation), a closed-loop red-blue teaming framework in which"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.05523","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/2606.05523/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":"2606.05523","created_at":"2026-06-05T01:14:53.727287+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.05523v1","created_at":"2026-06-05T01:14:53.727287+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.05523","created_at":"2026-06-05T01:14:53.727287+00:00"},{"alias_kind":"pith_short_12","alias_value":"EA3645LHD7CA","created_at":"2026-06-05T01:14:53.727287+00:00"},{"alias_kind":"pith_short_16","alias_value":"EA3645LHD7CA46XO","created_at":"2026-06-05T01:14:53.727287+00:00"},{"alias_kind":"pith_short_8","alias_value":"EA3645LH","created_at":"2026-06-05T01:14:53.727287+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/EA3645LHD7CA46XOL7T3OSJLRJ","json":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ.json","graph_json":"https://pith.science/api/pith-number/EA3645LHD7CA46XOL7T3OSJLRJ/graph.json","events_json":"https://pith.science/api/pith-number/EA3645LHD7CA46XOL7T3OSJLRJ/events.json","paper":"https://pith.science/paper/EA3645LH"},"agent_actions":{"view_html":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ","download_json":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ.json","view_paper":"https://pith.science/paper/EA3645LH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.05523&json=true","fetch_graph":"https://pith.science/api/pith-number/EA3645LHD7CA46XOL7T3OSJLRJ/graph.json","fetch_events":"https://pith.science/api/pith-number/EA3645LHD7CA46XOL7T3OSJLRJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ/action/storage_attestation","attest_author":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ/action/author_attestation","sign_citation":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ/action/citation_signature","submit_replication":"https://pith.science/pith/EA3645LHD7CA46XOL7T3OSJLRJ/action/replication_record"}},"created_at":"2026-06-05T01:14:53.727287+00:00","updated_at":"2026-06-05T01:14:53.727287+00:00"}