{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZXJZJAUWMXQJF2PXTUTNVGLEE6","short_pith_number":"pith:ZXJZJAUW","schema_version":"1.0","canonical_sha256":"cdd394829665e092e9f79d26da996427b9591a2fcc112c98c5936a19032de124","source":{"kind":"arxiv","id":"2508.17450","version":3},"attestation_state":"computed","paper":{"title":"Persuasion Dynamics in LLMs: Investigating Robustness and Adaptability in Knowledge and Safety with DuET-PD","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CY"],"primary_cat":"cs.CL","authors_text":"Bryan Chen Zhengyu Tan, Daniel Wai Kit Chin, Nancy F. Chen, Roy Ka-Wei Lee, Zhengyuan Liu","submitted_at":"2025-08-24T17:08:37Z","abstract_excerpt":"Large Language Models (LLMs) can struggle to balance gullibility to misinformation and resistance to valid corrections in persuasive dialogues, a critical challenge for reliable deployment. We introduce DuET-PD (Dual Evaluation for Trust in Persuasive Dialogues), a framework evaluating multi-turn stance-change dynamics across dual dimensions: persuasion type (corrective/misleading) and domain (knowledge via MMLU-Pro, and safety via SALAD-Bench). We find that even a state-of-the-art model like GPT-4o achieves only 27.32% accuracy in MMLU-Pro under sustained misleading persuasions. Moreover, res"},"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":"2508.17450","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-24T17:08:37Z","cross_cats_sorted":["cs.CY"],"title_canon_sha256":"9eedc0ae458f4016a77cb428b1094fe9a0ce76e84259c0c0753e6c5e3e1d9b2b","abstract_canon_sha256":"4f9669859d30553313c5d110aad4a74bfe8d8a43f74a3b4b844608e11ce25236"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:07:10.765698Z","signature_b64":"N71gFahaew1HMlcCK8Is7RVAI7sAra45EaReE5VaasxHPHXEcZMCZP6Bhgm2FKiTjwdee8b+z/0EfmchF0wsAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cdd394829665e092e9f79d26da996427b9591a2fcc112c98c5936a19032de124","last_reissued_at":"2026-07-05T12:07:10.765187Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:07:10.765187Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Persuasion Dynamics in LLMs: Investigating Robustness and Adaptability in Knowledge and Safety with DuET-PD","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CY"],"primary_cat":"cs.CL","authors_text":"Bryan Chen Zhengyu Tan, Daniel Wai Kit Chin, Nancy F. Chen, Roy Ka-Wei Lee, Zhengyuan Liu","submitted_at":"2025-08-24T17:08:37Z","abstract_excerpt":"Large Language Models (LLMs) can struggle to balance gullibility to misinformation and resistance to valid corrections in persuasive dialogues, a critical challenge for reliable deployment. We introduce DuET-PD (Dual Evaluation for Trust in Persuasive Dialogues), a framework evaluating multi-turn stance-change dynamics across dual dimensions: persuasion type (corrective/misleading) and domain (knowledge via MMLU-Pro, and safety via SALAD-Bench). We find that even a state-of-the-art model like GPT-4o achieves only 27.32% accuracy in MMLU-Pro under sustained misleading persuasions. Moreover, res"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.17450","kind":"arxiv","version":3},"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/2508.17450/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":"2508.17450","created_at":"2026-07-05T12:07:10.765245+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.17450v3","created_at":"2026-07-05T12:07:10.765245+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.17450","created_at":"2026-07-05T12:07:10.765245+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZXJZJAUWMXQJ","created_at":"2026-07-05T12:07:10.765245+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZXJZJAUWMXQJF2PX","created_at":"2026-07-05T12:07:10.765245+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZXJZJAUW","created_at":"2026-07-05T12:07:10.765245+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22633","citing_title":"Confident but Conflicted: Internal Uncertainty and Cognitive Dissonance Resolution in LLMs","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6","json":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6.json","graph_json":"https://pith.science/api/pith-number/ZXJZJAUWMXQJF2PXTUTNVGLEE6/graph.json","events_json":"https://pith.science/api/pith-number/ZXJZJAUWMXQJF2PXTUTNVGLEE6/events.json","paper":"https://pith.science/paper/ZXJZJAUW"},"agent_actions":{"view_html":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6","download_json":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6.json","view_paper":"https://pith.science/paper/ZXJZJAUW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.17450&json=true","fetch_graph":"https://pith.science/api/pith-number/ZXJZJAUWMXQJF2PXTUTNVGLEE6/graph.json","fetch_events":"https://pith.science/api/pith-number/ZXJZJAUWMXQJF2PXTUTNVGLEE6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6/action/storage_attestation","attest_author":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6/action/author_attestation","sign_citation":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6/action/citation_signature","submit_replication":"https://pith.science/pith/ZXJZJAUWMXQJF2PXTUTNVGLEE6/action/replication_record"}},"created_at":"2026-07-05T12:07:10.765245+00:00","updated_at":"2026-07-05T12:07:10.765245+00:00"}