Grammarly sued over AI feature that presented advice as coming from scientists without their consent

Grammarly shut down its Expert Review feature following a class action lawsuit. The feature presented AI-generated corrections as if they came from well-known authors and scientists — without their consent or knowledge. This is yet another case where an AI product crosses the line where a digital persona intersects with the rights of real people.

Grammarly is one of the most widely used tools for improving written text. According to the company, its services are used by more than 30 million people daily: students, journalists, managers, copywriters. The product started as a simple spell-checker, but over recent years has grown into a full-fledged AI platform with features for style, tone, and personalization.

Expert Review was one of such features. The idea seemed attractive: a user receives feedback not just from an algorithm, but in the style of specific recognized authors or scientists. In practice, however, according to the plaintiffs, the system did not simply draw inspiration from someone's style — it directly attributed advice to real people, using their names as a mark of quality.

Authors and academics whose names were involved knew nothing of such use, gave no consent, and received no compensation. The crux of the legal issue lies in the so-called right of publicity: a person's right to control the commercial use of their name, image, and reputation. In American law, this right is well protected, especially in states like California and New York.

The class action nature of the lawsuit means that those affected have united for a joint claim — and if the court accepts it as a single case, potential damages could be quite substantial. Grammarly responded quickly: on the same day the lawsuit became public, the feature was disabled. Such a move reduces current reputational damage and demonstrates the company's willingness to dialogue, however, it does not absolve responsibility for what has already occurred.

Legal experts note that prompt closure of a feature is typically taken into account by the court in favor of the defendant when determining damages, but does not exclude the responsibility itself. The Grammarly story is part of a broader wave. As AI products transition from general advice to personalized content, they increasingly rely on "real people" as a point of trust.

Celebrity deepfakes in advertising, AI voices of deceased musicians, chatbots imitating living influencers — the legal landscape around all this is still forming. Grammarly found itself in it unexpectedly: a text-editing feature seems far removed from the world of deepfakes. But the principle is the same — someone else's name is used as an asset without the permission of its owner.

For the market, this is a signal: even features that seem like harmless UX improvements carry legal risks if they are based on attribution to real people. Before using the name or image of any living person — an author, scientist, expert — as a "label" for AI advice, companies are obligated to obtain explicit written consent. Otherwise, the next lawsuit could prove more expensive than the savings from skipping a legal review.