{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:U66NK2GR5FDBBWAQRHCP3SWFW2","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":"3c2dd244e52523ed0fcacdc85723eeeec832bea9fb32f6f33156331be651cc1e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-24T04:25:59Z","title_canon_sha256":"8ce1c2c2eae0cb787d6472eec2a0f38054655abfc14b3e67d0623d347b5d9d3b"},"schema_version":"1.0","source":{"id":"2606.26161","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.26161","created_at":"2026-06-26T00:15:27Z"},{"alias_kind":"arxiv_version","alias_value":"2606.26161v1","created_at":"2026-06-26T00:15:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.26161","created_at":"2026-06-26T00:15:27Z"},{"alias_kind":"pith_short_12","alias_value":"U66NK2GR5FDB","created_at":"2026-06-26T00:15:27Z"},{"alias_kind":"pith_short_16","alias_value":"U66NK2GR5FDBBWAQ","created_at":"2026-06-26T00:15:27Z"},{"alias_kind":"pith_short_8","alias_value":"U66NK2GR","created_at":"2026-06-26T00:15:27Z"}],"graph_snapshots":[{"event_id":"sha256:52d927a3cb8e9c8122e9a112dfe6200cef75d1fd182275ff77e2799eb7d3673c","target":"graph","created_at":"2026-06-26T00:15:27Z","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/2606.26161/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Linear directions in activation space have been identified for both refusal and persona traits in instruction-tuned chat models, but the two have been studied as separate mechanisms. We show they interact: a compliant persona gates refusal. In Qwen2.5-7B-Instruct and Llama-3.1-8B-Instruct, we extract a compliant model-persona direction and a refusal direction and intervene on both. Compliant persona steering suppresses refusal -- in Llama, the refusal rate falls from 97% to 2%. Reintroducing the refusal direction partially restores refusal at late layers but not at early ones. Projecting out t","authors_text":"Qirui Li, Viola Zhong","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-24T04:25:59Z","title":"Refusal Lives Downstream of Persona in Chat Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.26161","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":"sha256:39f3cc04087cf77a6d5adcbea08c2563648c208c4eadab8fb4368315c2cf15e3","target":"record","created_at":"2026-06-26T00:15:27Z","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":"3c2dd244e52523ed0fcacdc85723eeeec832bea9fb32f6f33156331be651cc1e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-24T04:25:59Z","title_canon_sha256":"8ce1c2c2eae0cb787d6472eec2a0f38054655abfc14b3e67d0623d347b5d9d3b"},"schema_version":"1.0","source":{"id":"2606.26161","kind":"arxiv","version":1}},"canonical_sha256":"a7bcd568d1e94610d81089c4fdcac5b6891d5ba9b5d5946e5b094671c0a52db1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a7bcd568d1e94610d81089c4fdcac5b6891d5ba9b5d5946e5b094671c0a52db1","first_computed_at":"2026-06-26T00:15:27.628516Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-26T00:15:27.628516Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"whNKS68OICgrP1xm6LBrrD4ZtSYZslRkdmdnkTIO8I/cBB/qMQMiifSdyGvUX2xQd/F1zp7KLbcEeB3gebWdDQ==","signature_status":"signed_v1","signed_at":"2026-06-26T00:15:27.629013Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.26161","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:39f3cc04087cf77a6d5adcbea08c2563648c208c4eadab8fb4368315c2cf15e3","sha256:52d927a3cb8e9c8122e9a112dfe6200cef75d1fd182275ff77e2799eb7d3673c"],"state_sha256":"68e61977d61215d7c874493c4d789a8df1619bb40aecd3e9dd3ba43bbcdf87e1"}