Scientific journals are experiencing a surge in letters to the editor that appear to be generated by artificial intelligence, raising concerns among editors about the integrity of this important part of scholarly discourse. A recent study began when a researcher noticed suspicious letters referencing his own work inaccurately, leading to the discovery of a pattern: previously inactive authors suddenly publishing dozens or even hundreds of letters in a short time span. Editors report that these letters often arrive unusually quickly after a paper's publication and cover a wide range of topics, suggesting automated assistance.

The trend is driven in part by the incentive to boost academic credentials, as published letters are counted alongside full articles in scientific databases. While some suggest removing letters from these databases to deter abuse, editors warn this would also undermine valuable scientific discussion and critique. With no clear solution in sight, the proliferation of chatbot-generated correspondence poses a growing challenge to the credibility and usefulness of letters to the editor in scientific publishing.