Grammarly just launched a feature that's raising serious questions about AI consent and impersonation. The writing assistant's new 'expert review' tool claims to improve user writing with feedback from 'the world's great writers and thinkers' - but those experts never agreed to participate. Instead, the company is using AI to simulate their writing styles and voices, including tech journalists, to generate automated critiques. The move puts Grammarly at the center of a growing debate about whether AI companies can ethically create digital replicas of real people without permission.
Grammarly thinks it's found a shortcut to expert writing advice - just simulate the experts themselves. The company's recently-added feature promises users feedback styled after renowned writers and thinkers, but there's a catch that's hard to ignore: those experts never signed up for this.
The feature works by analyzing user text and generating critiques that supposedly mirror how specific writers would respond. Tech journalists are among those being simulated, their distinctive voices and editorial perspectives reduced to algorithmic outputs. Users can apparently request feedback 'from' particular writers, receiving AI-generated suggestions that imitate each person's style and approach to writing.
What makes this different from generic AI writing assistance is the explicit invocation of real people's identities. Grammarly isn't just offering 'journalist-style feedback' or 'literary analysis' - it's attaching actual names and reputations to AI-generated content. That distinction is fueling concerns among the writing community about consent, attribution, and the boundaries of AI impersonation.
The timing couldn't be more fraught. AI companies are already facing mounting criticism for training models on copyrighted content without permission. But this takes the controversy a step further - from using someone's published work as training data to actively deploying their persona as a product feature. It's one thing to learn from a journalist's articles; it's another to present AI output as if that journalist themselves reviewed your work.
