{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:BDK5HC7OFU22BBJNWPIT2KGR3Y","short_pith_number":"pith:BDK5HC7O","schema_version":"1.0","canonical_sha256":"08d5d38bee2d35a0852db3d13d28d1de399483807b9b6f9b03d9dddee605f561","source":{"kind":"arxiv","id":"2605.16312","version":1},"attestation_state":"computed","paper":{"title":"When Actions Disappear: Adversarial Action Removal in Self-Play Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Arahan Kujur","submitted_at":"2026-05-04T14:05:39Z","abstract_excerpt":"We study adversarial action masking in self-play reinforcement learning: an attacker selectively removes legal actions from a victim's action set. Unlike observation or action perturbations, removal eliminates decision options before the agent acts. Across poker games scaling from 6 to 5,531 information states and two non-poker domains, learned masking causes substantially more damage than random masking and learned perturbation baselines. The attack persists across Q-learning, PPO, NFSP, neural NFSP, and DQN victims; transfers across agents; is amplified by self-play; and shows no recovery un"},"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":"2605.16312","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-04T14:05:39Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"80ecfccb5f2452e33b84923566c22da2265a873fbaf68f5a9f4f66ec5ad7c8b1","abstract_canon_sha256":"f4d3f5e8ff05fcd72738ba0511ce9713a8fb3f88e12bb7a888c4b657e3553db0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:16.531727Z","signature_b64":"oD/vOMd6B0Av1yCXpmIlFsf4YRi3KFQKVIGCydXtkDDlR5xPg+oo3xkk7Zf0WHoE368e+kOu1IOgfP3gXWC6CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08d5d38bee2d35a0852db3d13d28d1de399483807b9b6f9b03d9dddee605f561","last_reissued_at":"2026-05-20T00:02:16.531146Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:16.531146Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When Actions Disappear: Adversarial Action Removal in Self-Play Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Arahan Kujur","submitted_at":"2026-05-04T14:05:39Z","abstract_excerpt":"We study adversarial action masking in self-play reinforcement learning: an attacker selectively removes legal actions from a victim's action set. Unlike observation or action perturbations, removal eliminates decision options before the agent acts. Across poker games scaling from 6 to 5,531 information states and two non-poker domains, learned masking causes substantially more damage than random masking and learned perturbation baselines. The attack persists across Q-learning, PPO, NFSP, neural NFSP, and DQN victims; transfers across agents; is amplified by self-play; and shows no recovery un"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.16312","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/2605.16312/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":"2605.16312","created_at":"2026-05-20T00:02:16.531239+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.16312v1","created_at":"2026-05-20T00:02:16.531239+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.16312","created_at":"2026-05-20T00:02:16.531239+00:00"},{"alias_kind":"pith_short_12","alias_value":"BDK5HC7OFU22","created_at":"2026-05-20T00:02:16.531239+00:00"},{"alias_kind":"pith_short_16","alias_value":"BDK5HC7OFU22BBJN","created_at":"2026-05-20T00:02:16.531239+00:00"},{"alias_kind":"pith_short_8","alias_value":"BDK5HC7O","created_at":"2026-05-20T00:02:16.531239+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/BDK5HC7OFU22BBJNWPIT2KGR3Y","json":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y.json","graph_json":"https://pith.science/api/pith-number/BDK5HC7OFU22BBJNWPIT2KGR3Y/graph.json","events_json":"https://pith.science/api/pith-number/BDK5HC7OFU22BBJNWPIT2KGR3Y/events.json","paper":"https://pith.science/paper/BDK5HC7O"},"agent_actions":{"view_html":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y","download_json":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y.json","view_paper":"https://pith.science/paper/BDK5HC7O","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.16312&json=true","fetch_graph":"https://pith.science/api/pith-number/BDK5HC7OFU22BBJNWPIT2KGR3Y/graph.json","fetch_events":"https://pith.science/api/pith-number/BDK5HC7OFU22BBJNWPIT2KGR3Y/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y/action/storage_attestation","attest_author":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y/action/author_attestation","sign_citation":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y/action/citation_signature","submit_replication":"https://pith.science/pith/BDK5HC7OFU22BBJNWPIT2KGR3Y/action/replication_record"}},"created_at":"2026-05-20T00:02:16.531239+00:00","updated_at":"2026-05-20T00:02:16.531239+00:00"}