Grammarly forced to remove AI feature after anger from writers and researchers

For several months now, AI assistants have been adding numerous features designed to improve writing, whether it's to correct grammar, suggest reformulations, or propose ideas. But these innovations can sometimes provoke strong reactions…

Indeed, that's precisely what just happened to Grammarly, one of the most widely used writing assistants in the world. The company recently had to disable an AI feature called "Expert Review, after a wave of criticism from writers, journalists, and researchers whose names were being used by the tool without permission.

An AI that mimicked the advice of real people

The feature is no longer accessible – Source: Grammarly

Specifically, the "Expert Review" feature had been introduced to help users improve their writing. The tool offered comments and writing suggestions presented as inspired by experts in the relevant field. These "experts" could be famous authors, scientists, or journalists. In some cases, the AI even presented remarks supposedly from figures such as bestselling novelists, academics, or technology specialists. The problem, quickly identified by users, was that these individuals had not given their consent to appear in the tool. The names of writers or researchers, sometimes deceased, were thus associated with comments automatically generated by an AI model.

A controversy that led to a swift suspension

As reported by Wired, in the face of criticism, the reaction was swift. Some authors denounced the misuse of their identities, while journalists and academics felt the tool could create the illusion of endorsement or collaboration where none existed. American journalist Julia Angwin even filed a class-action lawsuit against the company, arguing that using real names without authorization could constitute identity theft. Meanwhile, Shishir Mehrotra, CEO of Superhuman, the company that now owns Grammarly, announced the suspension of the feature. In a message posted on LinkedIn, he acknowledges that the criticisms made by experts were legitimate and that the company needed to revise its approach.

The company now explains that it wants to rethink the feature to allow experts to control how their name or work could be represented, or even to refuse this use altogether.

A new episode in the debate surrounding generative AI

This controversy illustrates the growing tensions surrounding generative AI models, where many tools rely on the analysis of vast datasets from the web, this which regularly raises questions related to consent, copyright, or the use of public figures' identities.

In the case of Grammarly, the situation highlights the difficulty of offering features inspired by human experts without giving the impression that they are actually participating in the service.

The suspension of "Expert Review" could thus mark a turning point in how AI tool developers design their features, so much so that for companies in the sector, the issue of representation and consent of creators is now becoming an increasingly sensitive subject…