An AI faked a reference in one of my papers. An editor caught it before publication. So I scanned 2.5 million biomedical papers. We found 4,046 citations to studies that don't exist. Up 12× since 2023. 98.4% uncorrected. Today in @TheLancet.
thelancet.com/journals/lance…
thelancet.com/journals/lance…