Grammarly disabled an AI review feature that cloned editors' style without their knowledge

Grammarly Disables AI Review Feature That Cloned Editors' Styles Without Their Knowledge

Superhuman, which owns Grammarly, disabled the Expert Review AI feature in March 2026—it had signed off on proposed text edits with the names of real journalists and editors, without asking their permission. As The Verge reports, among those affected were The Verge's editor-in-chief Nilay Patel and other employees of the publication.

How the Expert Review Feature Worked

Expert Review embedded comments from Grammarly into documents, saying that the proposed edit was "inspired" by the style of a particular well-known author. The Verge published a screenshot of a draft of their own article in Google Docs where a Grammarly AI comment was signed with the name of Nilay Patel—the editor never gave permission to use his name and style in this feature. The screenshot became direct proof that the names of real journalists and editors ended up in the product interface without any negotiations with them—Grammarly did not ask Patel himself or other Verge employees whose names appeared in the service's suggestions.

• Superhuman, which owns Grammarly, disabled the Expert Review feature
• Among the "experts" mentioned without permission were The Verge editor-in-chief Nilay Patel and other employees of the publication
• Aelian Gan, Superhuman's director of product management, made the statement about the feature's disabling
• The decision was made after The Verge previously exposed Grammarly's practice of using others' names in this feature

What Superhuman Said About Its Mistake

Superhuman acknowledged that it had misread user reactions to the feature and promised to rethink Expert Review from scratch—with real expert control over how (and whether) their names are used in the service.

"After careful consideration, we have decided to disable

Expert Review while we reconsider this feature to make it more useful for users and give experts real control over how they are—or aren't—represented," Aelian Gan, Superhuman's director of product management, said in a statement to The Verge.

"Based on the feedback we received, it's clear we missed the mark.

We apologize, and we will do things differently going forward," Gan added.

The feature was disabled only after The Verge first described how Expert Review worked and published a screenshot with Nilay Patel's name as evidence. Before this conflict with the publication, Superhuman had not disabled the feature or publicly commented on it—Gan's statement became the company's first official response to the claims.

What This Means

Grammarly is a popular AI text checking and editing service used by, among others, journalists and editors at major publications. The Expert Review story shows how quickly a product built on data about real people without their consent turns from a marketing move into a reputation risk: the company not only had to apologize but completely shut down the feature and rethink it from scratch. For other AI services that use others' names and styles for marketing purposes, this is a signal: expert consent now becomes a mandatory requirement, not a formality.