Language models caught 53% to 96% of planted trial errors; human reviewers caught 27%
A preprint tested 15 large language models on clinical trial reports with inserted errors, from an earlier experiment where untrained human reviewers scored 27% on major errors. The models scored 53% to 96%, and 58% to 99% on 58 trials published in 2026. They also flagged 14% to 29% of error types in control reports with no inserted errors. The preprint is not peer reviewed.