{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:FXSPYSGK4PSBLI4G62QLT5VJJV","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a5d6e8c5ac05ddedb6f684fb2ef5e57bf138762b969913d36d8758b910be7e69","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-08T16:21:36Z","title_canon_sha256":"335605a47ccb43370886ca71509a5347e0b22bb3be3413fc87ed97d1b1a15a64"},"schema_version":"1.0","source":{"id":"2606.09701","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.09701","created_at":"2026-06-09T02:09:04Z"},{"alias_kind":"arxiv_version","alias_value":"2606.09701v1","created_at":"2026-06-09T02:09:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09701","created_at":"2026-06-09T02:09:04Z"},{"alias_kind":"pith_short_12","alias_value":"FXSPYSGK4PSB","created_at":"2026-06-09T02:09:04Z"},{"alias_kind":"pith_short_16","alias_value":"FXSPYSGK4PSBLI4G","created_at":"2026-06-09T02:09:04Z"},{"alias_kind":"pith_short_8","alias_value":"FXSPYSGK","created_at":"2026-06-09T02:09:04Z"}],"graph_snapshots":[{"event_id":"sha256:bf2033d7b110b4d9d28de98cf40af9653f0d058cdff2b22204c914824307a63f","target":"graph","created_at":"2026-06-09T02:09:04Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2606.09701/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"AI red teaming must continually adapt to evolving attackers and defenders. Reinforcement learning offers a promising approach to discovering novel attacks, and co-training methods can produce more robust defenders in tandem. Recent works have demonstrated the efficacy of attacker-defender co-training by applying PPO and DPO, but report that GRPO is unstable in this setting. We introduce AdvGRPO, a co-training framework that makes GRPO viable for joint attacker-defender optimization using dense multi-channel rewards and decoupled advantage normalization. Training progresses through a curriculum","authors_text":"Amanda Minnich, Blake Bullwinkel, Eugenia Kim, Mark Russinovich","cross_cats":["cs.AI","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-08T16:21:36Z","title":"Learning to Attack and Defend: Adaptive Red Teaming of Language Models via GRPO"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09701","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:63ff41da71ab50e1a7824dc81b62577d9685d9771983de196f9b8386cee33da9","target":"record","created_at":"2026-06-09T02:09:04Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"a5d6e8c5ac05ddedb6f684fb2ef5e57bf138762b969913d36d8758b910be7e69","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-08T16:21:36Z","title_canon_sha256":"335605a47ccb43370886ca71509a5347e0b22bb3be3413fc87ed97d1b1a15a64"},"schema_version":"1.0","source":{"id":"2606.09701","kind":"arxiv","version":1}},"canonical_sha256":"2de4fc48cae3e415a386f6a0b9f6a94d73d11d1a57656bee8b92849b00656c77","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2de4fc48cae3e415a386f6a0b9f6a94d73d11d1a57656bee8b92849b00656c77","first_computed_at":"2026-06-09T02:09:04.664889Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T02:09:04.664889Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"gGnuOU7csLpiqmtA6DTR8E0Rt3vBpxJlwlC1DdTZBE+lr4it2Myv6E8JkO2UcNvbxCL84eLNmi+nzZAyBV1fDQ==","signature_status":"signed_v1","signed_at":"2026-06-09T02:09:04.671779Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.09701","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:63ff41da71ab50e1a7824dc81b62577d9685d9771983de196f9b8386cee33da9","sha256:bf2033d7b110b4d9d28de98cf40af9653f0d058cdff2b22204c914824307a63f"],"state_sha256":"6b86bb9753384f9b162c9918116d4b86bccaf9b986d8e6a7ee3f3ec235fcd4d1"}