Grammarly pulls AI feature that imitated famous authors after backlash and class-action lawsuit

Grammarly has disabled the controversial Expert Review feature, which generated edits in the manner of famous writers and scientists. The decision came after sharp criticism and amid a multimillion-dollar class action over the use of real names without consent.

How the mode worked

Expert Review offered users comments and editorial notes that appeared to be written with a specific author's or researcher's manner in mind. Instead of a neutral check of the text, the service imitated the tone and approach of recognizable public figures, turning ordinary editing into a stylized consultation.

That became the main point of conflict: the product did not simply help improve wording, it created the impression that a real person's intellectual persona stood behind the advice. The reference points included authors and scientists with long-established public voices.

For marketing, this looked impressive. Legally and ethically, it proved far riskier, because the platform tied AI-generated content to specific people without showing that they took part in the product or approved of such a format. A user could easily take it for personal expertise rather than a stylization assembled by a model from public speech patterns.

• Recommendations in the spirit of Stephen King
• Comments inspired by Neil deGrasse Tyson
• Hints referencing Carl Sagan
• A format that creates the feeling of an "expert review"

Why it ended up in court

The complaint against Grammarly is not about the use of generative AI as such, but about how the feature was packaged. The plaintiffs argue that the company used real names and recognizable reputations without permission, and may therefore have violated rights to name, likeness and commercial identity.

When a service sells a subscription and at the same time promises editing inspired by specific famous people, it looks less like impersonal automation and more like monetizing someone else's fame.

Against this backdrop, the class action looks like a logical continuation of the public reaction. Discontent around AI services increasingly arises where companies take a human style, voice or face and turn it into a product without direct consent. The Expert Review story hits exactly that nerve: users are offered a convenient feature, but the price of convenience is a blurred line between training on texts and commercial appropriation of a person's identity.

What the shutdown changes

Disabling Expert Review is not just a quick rollback of one failed feature. For Grammarly, it is a signal that the market is no longer ready to accept, without questions, AI tools that pose as living experts. That is especially true for a writing product, where trust in authorship, tone and editorial intervention is directly tied to who gives the advice and on what basis in each case.

Companies will most likely have to word such modes more carefully: not "as a famous author would say," but "in such-and-such a genre" or "with such-and-such a level of strictness." The difference seems cosmetic, but legally it is enormous. The more precisely a product ties a model's output to a specific person, the higher the risk of getting not only criticism on social media but a real lawsuit with costly consequences for the business and the brand.

What it means

For the AI market, this is one more marker: stylization after real people is ceasing to be a harmless feature and becoming a zone of direct legal risk. Users want convenient writing tools, but companies will have to clearly separate the model's help from another person's identity, name and reputation.

After the Grammarly case, any editorial AI service will have to choose its wording more carefully, not only its models and features, but also the copy on landing pages, in pricing plans and inside the interface.