Risk Evaluation · Study

Large language models can effectively convince people to believe conspiracies

NUWA research note · independent review pending

First published: 2026-01-08Source checked: 2026-10-25

Original source ↗Cite this record ↓Disinformation →

NUWA Research Note

Research question Design and assumptions Source findings NUWA interpretation Limits and contrary evidence Research and decision relevance Institutions and roles Original resources

Model-assisted reading · Selected sections of the original full text or official article; a complete section-by-section audit remains pending.

At a glance

NUWA interpretation: persuasive impact can favor false claims under the tested instructions; study versions must remain separate.

Research question

Can conversational AI increase conspiracy belief, and can truth constraints mitigate it?

Design and assumptions

The newly retrieved arXiv v3 describes four randomized online studies, N=3,996 US adults, with pre/post belief and debriefing measures.

Study 1 belief changes and evaluations of GPT-4o — Figure 2 · Figure 2 · Study 1 analytic sample, N = 1,092. Model-estimated belief changes, AI evaluations and trust under the reported study conditions. Original figure and source ↗

Source findings

Standard GPT-4o in Study 2 increased belief by an adjusted 11.9 points and decreased it by 12.9 in the opposite condition.

The earlier cached FAR abstract instead describes three studies, N=2,724; these versions have different denominators.

NUWA interpretation

NUWA interpretation: persuasive impact can favor false claims under the tested instructions; study versions must remain separate.

Limits and contrary evidence

US online samples and immediate self-report outcomes limit population and long-term inference; cross-study comparisons are not all randomized contrasts.

Research and decision relevance

Model recommendation: cite arXiv:2601.05050v3 explicitly and preserve truth-constraint and debriefing conditions.

Precise source locations

Large language models can effectively convince people to believe conspiracies · arXiv:2601.05050v3 (15 July 2026); Methods: Overview of study designsSelected original resource https://arxiv.org/html/2601.05050v3; sha256:506938d68312ef63bdfbf4af97aa957f36c05f92299ff1032732f0ecc2bc8f12; archived source capture; exact capture time not independently established
Large language models can effectively convince people to believe conspiracies · Methods: Participants, recruitment and sample sizesSelected original resource https://arxiv.org/html/2601.05050v3; sha256:506938d68312ef63bdfbf4af97aa957f36c05f92299ff1032732f0ecc2bc8f12; archived source capture; exact capture time not independently established
Large language models can effectively convince people to believe conspiracies · Methods: Measures and outcomesSelected original resource https://arxiv.org/html/2601.05050v3; sha256:506938d68312ef63bdfbf4af97aa957f36c05f92299ff1032732f0ecc2bc8f12; archived source capture; exact capture time not independently established
Large language models can effectively convince people to believe conspiracies · Results: Can design interventions mitigate the bunking effect?Selected original resource https://arxiv.org/html/2601.05050v3; sha256:506938d68312ef63bdfbf4af97aa957f36c05f92299ff1032732f0ecc2bc8f12; archived source capture; exact capture time not independently established
Large language models can effectively convince people to believe conspiracies · DiscussionSelected original resource https://arxiv.org/html/2601.05050v3; sha256:506938d68312ef63bdfbf4af97aa957f36c05f92299ff1032732f0ecc2bc8f12; archived source capture; exact capture time not independently established

Source checking, extraction, editorial approval and replication are recorded separately. Methodology →

Institutions and roles

Published by FAR.AI

Evaluation protocols

Conspiracy belief experiments →

Cite and reuse

Cite the original publication for its findings. Cite this NUWA record separately for its explanation and retained snapshot.

BibTeX ↓RIS ↓Record + source metadata · JSON ↓Reading coverage · JSON Suggest a correction

Original publication · BibTeX

@misc{conspiracy-persuasion,
  title = {Large language models can effectively convince people to believe conspiracies},
  url = {https://www.far.ai/research/large-language-models-can-effectively-convince-people-to-believe-conspiracies},
  year = {2026},
  date = {2026-01-08},
  version = {FAR.AI report 2026-01-08},
  note = {FAR.AI report 2026-01-08}
}

NUWA record · BibTeX

@misc{conspiracy-persuasion-nuwa,
  title = {Large language models can effectively convince people to believe conspiracies — NUWA research record},
  url = {https://whitzard.tech/en/nuwa/whitzard-index/work/conspiracy-persuasion/},
  version = {2026-10-29.1},
  organization = {NUWA Frontier AI Safety Lab},
  howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--conspiracy-persuasion.json},
  note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--conspiracy-persuasion.json.}
}

Evidence snapshot 2026-10-29.1 · Index updates · RSS

References

[1] Large language models can effectively convince people to believe conspiracies. 2026-01-08. v3 ; Official publication masthead and article byline; arXiv:2601.05050v3 (15 July 2026); Methods: Overview of study designs; Methods: Participants, recruitment and sample sizes