AI-generated misinformation rises in fact-checking investigations

A study conducted by Full Fact and the University of Westminster reveals a significant rise in AI-generated misinformation. Between January 2025 and March 2026, confirmed or suspected AI content accounted for 24% of all fact-checks investigated, a substantial increase from 8% the previous year.

Researchers analyzed 112 pieces of AI-generated or altered content that reached tens of millions of viewers. The study noted that as generative tools improve, traditional visual cues of synthetic media, such as anatomical errors, are becoming less frequent. Audio-only fakes were identified as particularly difficult to verify.

The research utilized a harm-risk assessment model to evaluate the impact of this content. The findings indicate that 83.9% of the analyzed entries created substantively false or misleading understandings, contributing to a broad erosion of public trust.