{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XWXF2MK2L2V7BVXMVBVCG43UWB","short_pith_number":"pith:XWXF2MK2","schema_version":"1.0","canonical_sha256":"bdae5d315a5eabf0d6eca86a237374b048a0af19989895c34e4e94fda65484c0","source":{"kind":"arxiv","id":"2605.21006","version":1},"attestation_state":"computed","paper":{"title":"Playing Devil's Advocate: Off-the-Shelf Persona Vectors Rival Targeted Steering for Sycophancy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Ishaan Kelkar, Madhur Panwar, Maheep Chaudhary, Nebras Alam, Vasu Sharma, Vikram Kakaria","submitted_at":"2026-05-20T10:43:17Z","abstract_excerpt":"We study the effect of different persona on \\textbf{sycophancy}: model's agreement with users even when the user is incorrect. The standard mitigation, Contrastive Activation Addition (CAA), derives a steering direction from labelled pairs of sycophantic and honest responses. This study evaluates whether off-the-shelf persona steering vectors, originally developed for general role-playing and not trained on sycophancy data, can serve as an alternative. In two instruction-tuned models, steering toward personas characterised by doubt or scrutiny reduces sycophancy to approximately $68\\%$ and $98"},"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":"2605.21006","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-20T10:43:17Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"2b02046cce0d121366f204a85355fcb59644ad42a1bb66a115ddae54af837dcb","abstract_canon_sha256":"627a611ae08ac9d28eb54424d01db7967b2fe83cb55e4ece0863da270c0e0d1c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:31.997715Z","signature_b64":"mBita2WL4y5q7rEg+O0u92ZxD6rwT9M4bJQrXzH4iWfylZAcmBulQsGcMaQ88/JP3+IyJULXDwgXdOBIGunqDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bdae5d315a5eabf0d6eca86a237374b048a0af19989895c34e4e94fda65484c0","last_reissued_at":"2026-05-21T01:05:31.997228Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:31.997228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Playing Devil's Advocate: Off-the-Shelf Persona Vectors Rival Targeted Steering for Sycophancy","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Ishaan Kelkar, Madhur Panwar, Maheep Chaudhary, Nebras Alam, Vasu Sharma, Vikram Kakaria","submitted_at":"2026-05-20T10:43:17Z","abstract_excerpt":"We study the effect of different persona on \\textbf{sycophancy}: model's agreement with users even when the user is incorrect. The standard mitigation, Contrastive Activation Addition (CAA), derives a steering direction from labelled pairs of sycophantic and honest responses. This study evaluates whether off-the-shelf persona steering vectors, originally developed for general role-playing and not trained on sycophancy data, can serve as an alternative. In two instruction-tuned models, steering toward personas characterised by doubt or scrutiny reduces sycophancy to approximately $68\\%$ and $98"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.21006","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/2605.21006/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":"2605.21006","created_at":"2026-05-21T01:05:31.997304+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.21006v1","created_at":"2026-05-21T01:05:31.997304+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.21006","created_at":"2026-05-21T01:05:31.997304+00:00"},{"alias_kind":"pith_short_12","alias_value":"XWXF2MK2L2V7","created_at":"2026-05-21T01:05:31.997304+00:00"},{"alias_kind":"pith_short_16","alias_value":"XWXF2MK2L2V7BVXM","created_at":"2026-05-21T01:05:31.997304+00:00"},{"alias_kind":"pith_short_8","alias_value":"XWXF2MK2","created_at":"2026-05-21T01:05:31.997304+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/XWXF2MK2L2V7BVXMVBVCG43UWB","json":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB.json","graph_json":"https://pith.science/api/pith-number/XWXF2MK2L2V7BVXMVBVCG43UWB/graph.json","events_json":"https://pith.science/api/pith-number/XWXF2MK2L2V7BVXMVBVCG43UWB/events.json","paper":"https://pith.science/paper/XWXF2MK2"},"agent_actions":{"view_html":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB","download_json":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB.json","view_paper":"https://pith.science/paper/XWXF2MK2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.21006&json=true","fetch_graph":"https://pith.science/api/pith-number/XWXF2MK2L2V7BVXMVBVCG43UWB/graph.json","fetch_events":"https://pith.science/api/pith-number/XWXF2MK2L2V7BVXMVBVCG43UWB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB/action/storage_attestation","attest_author":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB/action/author_attestation","sign_citation":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB/action/citation_signature","submit_replication":"https://pith.science/pith/XWXF2MK2L2V7BVXMVBVCG43UWB/action/replication_record"}},"created_at":"2026-05-21T01:05:31.997304+00:00","updated_at":"2026-05-21T01:05:31.997304+00:00"}