Grammarly has shut down its "Expert Review" AI feature after it surfaced that the tool was presenting AI-generated edit suggestions as being "inspired by" real, named writers - without those writers ever agreeing to participate.
The feature worked like this: when users asked Grammarly to review their writing, the tool would attribute its suggestions to specific human experts, implying their editorial judgment shaped the feedback. Among those named were journalists at The Verge, including the publication's editor-in-chief. None of them had been contacted, compensated, or asked for permission.
Grammarly confirmed the decision in a statement, saying it would "reimagine the feature to make it more useful for users, while giving experts real control." The company framed the shutdown as a design rethink rather than an admission of wrongdoing, but the timeline speaks for itself: the feature disappeared shortly after journalists publicly flagged the issue.
This lands in a growing pile of cases where AI companies have treated real people's work, names, or likenesses as raw material. The difference here is the specificity. Grammarly wasn't just training on public writing - it was attaching individual names to AI output, creating the impression of endorsement. That crosses a line from "we trained on the internet" into something closer to unauthorized use of someone's professional identity.
For Grammarly's 30+ million daily users, the practical impact is minimal. Expert Review was a relatively new addition, and the core grammar and style tools remain unchanged. But the episode raises a question worth thinking about as more AI tools add "expert" or "personality" layers: whose voice is actually behind the suggestions, and did they agree to be there?
Grammarly says a redesigned version with proper expert controls is coming. No timeline was given.