{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,16]],"date-time":"2026-10-16T04:43:49Z","timestamp":1792125829675,"version":"4.4.0"},"reference-count":45,"publisher":"Oxford University Press (OUP)","issue":"10","license":[{"start":{"date-parts":[[2026,10,9]],"date-time":"2026-10-09T00:00:00Z","timestamp":1791504000000},"content-version":"vor","delay-in-days":9,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100010663","name":"European Research Council","doi-asserted-by":"publisher","award":["852334"],"award-info":[{"award-number":["852334"]}],"id":[{"id":"10.13039\/100010663","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,9,30]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Conversational AI has been proposed as a scalable way to correct public misconceptions, yet its effectiveness may depend on perceptions of political neutrality. As large language models become objects of partisan conflict, political elites increasingly portray them as ideologically aligned. We test whether such credibility attacks reduce persuasion by conversational AI. In a preregistered US survey experiment (N = 2,144), participants completed a three-round conversation with ChatGPT (GPT-4.1) about an economic policy misconception they personally held. The conversation corrected misconceptions across all conditions, but a short message indicating that the model was biased against the respondent's political party attenuated this correction by 28% relative to a neutral control and reduced trust in and willingness to use AI. Even under this adversarial framing, the majority of the persuasion effect survived. Pairwise transcript comparisons suggest a behavioral mechanism: treated respondents pushed back more and engaged less receptively rather than disengaging. These findings identify perceived partisan alignment as a user-side lever that conditions, but does not eliminate, the persuasive power of conversational AI.<\/jats:p>","DOI":"10.1093\/pnasnexus\/pgag324","type":"journal-article","created":{"date-parts":[[2026,10,9]],"date-time":"2026-10-09T11:25:41Z","timestamp":1791545141000},"source":"Crossref","is-referenced-by-count":0,"title":["Perceived political bias in large language models reduces persuasive abilities"],"prefix":"10.1093","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0823-0436","authenticated-orcid":false,"given":"Matthew","family":"DiGiuseppe","sequence":"first","affiliation":[{"name":"Institute of Political Science, Leiden University , Wassenaarseweg 52, Leiden 2333 AK, South Holland ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7472-2247","authenticated-orcid":false,"given":"Joshua","family":"Robison","sequence":"additional","affiliation":[{"name":"Institute of Political Science, Leiden University , Wassenaarseweg 52, Leiden 2333 AK, South Holland ,","place":["The Netherlands"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,10,9]]},"reference":[{"key":"2026101523470805600_pgag324-B1","doi-asserted-by":"crossref","first-page":"eadq1814","DOI":"10.1126\/science.adq1814","article-title":"Durably reducing conspiracy beliefs through dialogues with AI","volume":"385","author":"Costello","year":"2024","journal-title":"Science"},{"key":"2026101523470805600_pgag324-B2","doi-asserted-by":"crossref","first-page":"e2403116121","DOI":"10.1073\/pnas.2403116121","article-title":"Evaluating the persuasive influence of political microtargeting with large language models","volume":"121","author":"Hackenburg","year":"2024","journal-title":"Proc Natl Acad Sci U S A."},{"key":"2026101523470805600_pgag324-B3","author":"Noels"},{"key":"2026101523470805600_pgag324-B4","doi-asserted-by":"crossref","first-page":"eaea3884","DOI":"10.1126\/science.aea3884","article-title":"The levers of political persuasion with conversational artificial intelligence","volume":"390","author":"Hackenburg","year":"2025","journal-title":"Science"},{"key":"2026101523470805600_pgag324-B5","doi-asserted-by":"crossref","first-page":"e2412815122","DOI":"10.1073\/pnas.2412815122","article-title":"Testing theories of political persuasion using AI","volume":"122","author":"Argyle","year":"2025","journal-title":"Proc Natl Acad Sci U S A."},{"key":"2026101523470805600_pgag324-B6","doi-asserted-by":"crossref","first-page":"394","DOI":"10.1038\/s41586-025-09771-9","article-title":"Persuading voters using human\u2013artificial intelligence dialogues","volume":"648","author":"Lin","year":"2025","journal-title":"Nature"},{"key":"2026101523470805600_pgag324-B7","doi-asserted-by":"crossref","first-page":"6037","DOI":"10.1038\/s41467-025-61345-5","article-title":"LLM-generated messages can persuade humans on policy issues","volume":"16","author":"Bai","year":"2025","journal-title":"Nat Commun."},{"key":"2026101523470805600_pgag324-B8","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1177\/20531680251414926","article-title":"Deep canvassing with automated conversational agents: personalized messaging to change attitudes","volume":"13","author":"Offer-Westort","year":"2026","journal-title":"Res Politics"},{"key":"2026101523470805600_pgag324-B9","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3579592","article-title":"Working with AI to persuade: examining a large language model's ability to generate pro-vaccination messages","volume":"7","author":"Karinshak","year":"2023","journal-title":"Proc ACM Hum Comput Interact."},{"key":"2026101523470805600_pgag324-B10","author":"Goel","year":"2024"},{"key":"2026101523470805600_pgag324-B11","doi-asserted-by":"publisher","first-page":"pgaf325","DOI":"10.1093\/pnasnexus\/pgaf325","article-title":"Dialogues with large language models reduce conspiracy beliefs even when the AI is perceived as human","volume":"4","author":"Boissin","year":"2025","journal-title":"PNAS Nexus"},{"key":"2026101523470805600_pgag324-B12","doi-asserted-by":"crossref","first-page":"2420","DOI":"10.1038\/s41562-025-02309-z","article-title":"The impact of advanced AI systems on democracy","volume":"9","author":"Summerfield","year":"2025","journal-title":"Nat Hum Behav."},{"key":"2026101523470805600_pgag324-B13","author":"Czarnek","year":"2025"},{"key":"2026101523470805600_pgag324-B14","author":"Costello"},{"key":"2026101523470805600_pgag324-B15","doi-asserted-by":"crossref","first-page":"e0306621","DOI":"10.1371\/journal.pone.0306621","article-title":"The political preferences of LLMs","volume":"19","author":"Rozado","year":"2024","journal-title":"PLoS One"},{"key":"2026101523470805600_pgag324-B16","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1007\/s11127-023-01097-2","article-title":"More human than human: measuring ChatGPT political bias","volume":"198","author":"Motoki","year":"2024","journal-title":"Public Choice"},{"key":"2026101523470805600_pgag324-B17","volume-title":"Measuring perceived slant in large language models through user evaluations","author":"Westwood","year":"2025"},{"key":"2026101523470805600_pgag324-B18","doi-asserted-by":"crossref","first-page":"5224","DOI":"10.1038\/s41598-024-55686-2","article-title":"Bias of AI-generated content: an examination of news produced by large language models","volume":"14","author":"Fang","year":"2024","journal-title":"Sci Rep."},{"key":"2026101523470805600_pgag324-B19","volume-title":"Communication and persuasion: psychological studies of opinion change","author":"Hovland","year":"1953"},{"key":"2026101523470805600_pgag324-B20","doi-asserted-by":"crossref","DOI":"10.1017\/CBO9780511818691","volume-title":"The nature and origins of mass opinion","author":"Zaller","year":"1992"},{"key":"2026101523470805600_pgag324-B21","volume-title":"The democratic dilemma: can citizens learn what they need to know?","author":"Lupia","year":"1998"},{"key":"2026101523470805600_pgag324-B22","doi-asserted-by":"crossref","first-page":"1041","DOI":"10.1111\/0022-3816.00100","article-title":"On the limits of framing effects: who can frame?","volume":"63","author":"Druckman","year":"2001","journal-title":"J Polit."},{"key":"2026101523470805600_pgag324-B23","doi-asserted-by":"crossref","first-page":"243","DOI":"10.1111\/j.1559-1816.2004.tb02547.x","article-title":"The persuasiveness of source credibility: a critical review of five decades\u2019 evidence","volume":"34","author":"Pornpitakpan","year":"2004","journal-title":"J Appl Soc Psychol."},{"key":"2026101523470805600_pgag324-B24","doi-asserted-by":"crossref","first-page":"57","DOI":"10.1017\/S0003055412000500","article-title":"How elite partisan polarization affects public opinion formation","volume":"107","author":"Druckman","year":"2013","journal-title":"Am Polit Sci Rev."},{"key":"2026101523470805600_pgag324-B25","doi-asserted-by":"crossref","first-page":"1683","DOI":"10.1086\/717085","article-title":"Partisan influence in suspicious times","volume":"84","author":"Robison","year":"2022","journal-title":"J Polit."},{"key":"2026101523470805600_pgag324-B26","doi-asserted-by":"crossref","first-page":"e62","DOI":"10.1017\/S0007123424000899","article-title":"Evidence can change partisan minds but less so in hostile contexts","volume":"55","author":"Kim","year":"2025","journal-title":"Br J Polit Sci."},{"key":"2026101523470805600_pgag324-B27","doi-asserted-by":"publisher","DOI":"10.37016\/mr-2020-127","article-title":"Misinformation reloaded? Fears about the impact of generative AI on misinformation areoverblown","volume":"4","author":"Simon","year":"2023","journal-title":"Harv Kennedy Sch Misinformation Rev"},{"key":"2026101523470805600_pgag324-B28","doi-asserted-by":"crossref","first-page":"1032","DOI":"10.1037\/0022-3514.51.5.1032","article-title":"Central and peripheral routes to persuasion: an individual difference perspective","volume":"51","author":"Cacioppo","year":"1986","journal-title":"J Pers Soc Psychol."},{"key":"2026101523470805600_pgag324-B29","first-page":"137","volume-title":"The SAGE handbook of persuasion: developments in theory and practice","author":"O\u2019Keefe","year":"2013","edition":"2nd Ed"},{"key":"2026101523470805600_pgag324-B30","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1017\/pan.2026.10041","article-title":"Text as behavior","author":"Wasow","year":"2026","journal-title":"Polit Anal"},{"key":"2026101523470805600_pgag324-B31","doi-asserted-by":"crossref","first-page":"480","DOI":"10.1037\/0033-2909.108.3.480","article-title":"The case for motivated reasoning","volume":"108","author":"Kunda","year":"1990","journal-title":"Psychol Bull."},{"key":"2026101523470805600_pgag324-B32","doi-asserted-by":"crossref","first-page":"129","DOI":"10.1111\/pops.12164","article-title":"Political parties, motivated reasoning, and public opinion formation","volume":"35","author":"Leeper","year":"2014","journal-title":"Polit Psychol."},{"key":"2026101523470805600_pgag324-B33","first-page":"1","volume-title":"Emerging trends in the social and behavioral sciences","author":"Kahan","year":"2016"},{"key":"2026101523470805600_pgag324-B34","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1037\/0022-3514.71.1.5","article-title":"A disconfirmation bias in the evaluation of arguments","volume":"71","author":"Edwards","year":"1996","journal-title":"J Pers Soc Psychol."},{"key":"2026101523470805600_pgag324-B35","doi-asserted-by":"crossref","first-page":"755","DOI":"10.1111\/j.1540-5907.2006.00214.x","article-title":"Motivated skepticism in the evaluation of political beliefs","volume":"50","author":"Taber","year":"2006","journal-title":"Am J Pol Sci."},{"key":"2026101523470805600_pgag324-B36","doi-asserted-by":"crossref","first-page":"563","DOI":"10.1111\/j.1467-9221.2010.00772.x","article-title":"The affective tipping point: do motivated reasoners ever \u201cget it\u201d?","volume":"31","author":"Redlawsk","year":"2010","journal-title":"Polit Psychol."},{"key":"2026101523470805600_pgag324-B37","doi-asserted-by":"crossref","first-page":"630","DOI":"10.1093\/poq\/nfag013","article-title":"Scaling open-ended survey responses using LLM-paired comparisons","volume":"90","author":"DiGiuseppe","year":"2026","journal-title":"Public Opin Q."},{"key":"2026101523470805600_pgag324-B38","doi-asserted-by":"crossref","first-page":"581","DOI":"10.1093\/biomet\/63.3.581","article-title":"Inference and missing data","volume":"63","author":"Rubin","year":"1976","journal-title":"Biometrika"},{"key":"2026101523470805600_pgag324-B39","doi-asserted-by":"crossref","first-page":"824","DOI":"10.1111\/ajps.12432","article-title":"How getting the facts right can fuel partisan-motivated reasoning","volume":"63","author":"Bisgaard","year":"2019","journal-title":"Am J Pol Sci."},{"key":"2026101523470805600_pgag324-B40","doi-asserted-by":"crossref","first-page":"327","DOI":"10.1086\/683832","article-title":"Credible sources and sophisticated voters: when does new information induce economic voting?","volume":"78","author":"Alt","year":"2016","journal-title":"J Polit."},{"key":"2026101523470805600_pgag324-B41","doi-asserted-by":"crossref","first-page":"351","DOI":"10.1080\/10584609.2020.1772418","article-title":"When do partisans stop following the leader?","volume":"38","author":"Agadjanian","year":"2021","journal-title":"Polit Commun."},{"key":"2026101523470805600_pgag324-B42","doi-asserted-by":"crossref","first-page":"568","DOI":"10.1038\/s41562-023-01551-7","article-title":"Partisans\u2019 receptivity to persuasive messaging is undiminished by countervailing party leader cues","volume":"7","author":"Tappin","year":"2023","journal-title":"Nat Hum Behav."},{"key":"2026101523470805600_pgag324-B43","doi-asserted-by":"crossref","first-page":"112509","DOI":"10.1016\/j.econlet.2025.112509","article-title":"Evaluating online data collection platforms using a simple rule-following task","volume":"255","author":"Suri","year":"2025","journal-title":"Econ Lett."},{"key":"2026101523470805600_pgag324-B44","doi-asserted-by":"crossref","first-page":"106","DOI":"10.3758\/s13428-025-02618-1","article-title":"Evaluating mobile-based data collection for crowdsourcing behavioral research","volume":"57","author":"Esch","year":"2025","journal-title":"Behav Res Methods."},{"key":"2026101523470805600_pgag324-B45","doi-asserted-by":"crossref","first-page":"e2413443122","DOI":"10.1073\/pnas.2413443122","article-title":"Scaling language model size yields diminishing returns for single-message political persuasion","volume":"122","author":"Hackenburg","year":"2025","journal-title":"Proc Natl Acad Sci U S A."}],"container-title":["PNAS Nexus"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/pnasnexus\/article-pdf\/5\/10\/pgag324\/71786439\/pgag324.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/pnasnexus\/article-pdf\/5\/10\/pgag324\/71786439\/pgag324.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,16]],"date-time":"2026-10-16T03:47:18Z","timestamp":1792122438000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/pnasnexus\/article\/doi\/10.1093\/pnasnexus\/pgag324\/8898473"}},"subtitle":[],"editor":[{"given":"Michael","family":"Muthukrishna","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,9,30]]},"references-count":45,"journal-issue":{"issue":"10","published-print":{"date-parts":[[2026,9,30]]}},"URL":"https:\/\/doi.org\/10.1093\/pnasnexus\/pgag324","relation":{},"ISSN":["2752-6542"],"issn-type":[{"value":"2752-6542","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,10]]},"published":{"date-parts":[[2026,9,30]]},"article-number":"pgag324"}}