A Science Expression of Concern (June 2026) flags Costello, Pennycook & Rand's 2024 conspiracy-debunking paper for screening-criteria inconsistencies

conspiracy-beliefLLM-persuasionresearch-integrityretraction-watchexpression-of-concern

Science issued an Expression of Concern — reported by Retraction Watch on 2026-06-11 — for Costello, Pennycook & Rand's September 2024 paper "Durably Reducing Conspiracy Beliefs Through Dialogues With AI", a paper cited 192 times per Clarivate Web of Science (as of that report) and winner of the Newcomb Cleveland Prize of the American Association for the Advancement of Science — which publishes Science — awarded to the three authors in February 2026. The authors found their public dataset made some reported values hard to reproduce, and reported to Science that they had discovered "inconsistencies in the application of screening criteria between the manuscript and published analysis pipeline." They supplied updated raw data; their own reanalysis "produces results that match those in the original article in direction, statistical significance, and substantive size," but Science's own evaluation of the correction was still pending as of Retraction Watch's report. Status as of 2026-10-18 (scheduled audit): still pending. The authors posted a correction-and- reproducibility package — correction ledger, corrected article and supplementary figures, replacement supplementary materials, deidentified data, code, checksums — to Zenodo on 2026-09-27 (record 22754984); no published Science correction or retraction has been located, so the Expression of Concern (Science 392(6803):1131, doi 10.1126/science.aej2383) remains the operative notice.

This bears on the evidentiary base of the whole Costello/Pennycook/Rand conspiracy-persuasion research program, which also includes claim-llm-conspiracy-persuasion-is-dual-use, claim-2026-dual-use-paper-found-mirror-image-spillover-on-gcbs, and claim-2026-symmetry-result-answers-a-question-2024-paper-posed — three notes that, per the single-source concentration cap (sources.md), already max out what the vault can rest directly on the unrefereed 2026 preprint alone (see question-corroborate-costello-2026-arxiv-2601-05050-single-source-cap). The EOC does not touch the 2026 preprint's own data and does not discharge that cap; it is a separate reason for caution about the lineage's foundational paper. Thomas Costello is a co-author of both the flagged 2024 paper and its 2026 follow-on.