aceModerate evidence
Published recordMedia & ArtsProblem Space
More
Synthetic News, Natural Doubts? A Meta-Analysis of Credibility Perceptions of AI-Generated News
- Truvace record
- TRV-2026-2712raw data (JSON)
- Version
- v1Published
- First published
- Evidence type
- Systematic review or meta-analysis
- Sources
- 1 sourcePeer-reviewed
- sha256
- 8c88abd77e76…check it
As generative artificial intelligence (AI) becomes increasingly integrated into news production, concerns about the credibility of synthetic news content are intensifying. While some studies report that audiences find AI-generated news less credible than human-written content, others reveal mixed or even reversed findings. This meta-analysis synthesizes findings across 31 studies (41 effect sizes) to assess how the source label (AI vs. human) influences news users' evaluations of source credibility and message c…
The problem
A meta-analysis of 31 studies (41 effect sizes) found news labeled as AI-generated got a small but statistically significant credibility penalty, on both source and message credibility, compared with news labeled as human-written.
What this doesn’t fix
Two of the three moderators, topic value-ladenness and participants' source orientation, showed only descriptive trends and did not reach statistical significance, so the sources of variation between studies remain only partly explained.
Sources
1 on file- Cyberpsychology, Behavior, and Social Networkingdoi:10.1177/21522715261439452
Cite this record
v1Truvace Impact Record TRV-2026-2712, v1: “Synthetic News, Natural Doubts? A Meta-Analysis of Credibility Perceptions of AI-Generated News.” Truvace, 2026-11-12. /record/TRV-2026-2712 (accessed on the day you copy it). sha256 8c88abd77e7689a4…
Change history
1 versionEvery change made to this record since it was first published, and why. Nothing has changed since it went up.
Certified into the record
Check this record hasn’t been altered
How to check it yourself, without trusting this page
Download any version from /api/record/TRV-2026-2712 and hash its canonical field yourself, for example with shasum -a 256. You should get its content_hash. Each version’s text also ends with prev:and the hash of the version before it (version 1 points to 64 zeros). If anyone had changed a single character of any version after it was saved, those hashes wouldn’t line up.