Grammarly turned authors into "AI editors" without their consent, class-action lawsuit alleges

Investigative journalist Julia Angwin has filed a class action against Grammarly, accusing the company of unlawfully using authors' work to train AI tools. According to the plaintiffs, the service turned professional writers into so-called "AI editors" without their knowledge or consent, violating their rights to privacy and publicity.

The core of the complaint: Grammarly collected the texts of its users, including professional journalists and writers, and used them to train automatic editing algorithms. In effect, the AI learned to work with text the way human authors do, by analyzing their decisions, style and edits. Authors were offered no notification, no request for consent and no compensation of any kind.

The suit rests on two legal grounds: violation of the right to privacy and violation of the right to publicity. The latter protects a person from the commercial use of their identity, style or reputation without explicit permission.

Julia Angwin is a well-known American journalist, the winner of numerous professional awards and the founder of The Markup, a publication that specializes in investigations into technology, data and corporate surveillance. Her book "Dragnet Nation" became one of the defining texts of the era of digital privacy. Angwin is the lead plaintiff in the class action.

Thousands of other Grammarly users may stand behind her: all professional authors whose texts could have been used in a similar way without their knowledge.

The suit was filed in a US federal court. The plaintiffs seek not only damages but also a ban on further use of authors' data without explicit consent, as well as full transparency about exactly which materials were used to train the company's AI models. If the case is granted class-action status, hundreds of thousands of users of the service worldwide, who never consented to the use of their texts for such purposes, could fall under it.

Grammarly is one of the world's most popular AI tools for working with text: more than 30 million people use it to check grammar, improve style and generate content. The service is actively promoted to journalists, writers and professional editors, which makes the plaintiffs' claims especially damaging to the company's reputation. In recent years Grammarly has aggressively developed AI features, integrating large language models, and it was this expansion that drew the plaintiffs' attention.

The case is part of a broader wave of lawsuits against tech companies accused of using content without authorization to train AI. Similar claims have been brought against OpenAI, Stability AI, GitHub Copilot, Meta* and several other players. The Grammarly suit has a distinct emphasis, however: it is not simply about reproducing content, but about appropriating the editorial style of specific authors, that is, the commercial use of their professional identity.

The legal theory here is non-trivial. The right of publicity has historically applied to celebrities, protecting names, faces and voices from commercial exploitation without consent. But in the era of generative AI, lawyers are beginning to extend the concept: if a model learns to imitate an author's style, is that not the use of that person's intangible asset? If an AI is sold as a tool that "edits like a professional," who paid for that competence? The court's answer to these questions will set a precedent for the whole industry.

For authors and journalists the signal is clear: consent to data processing in a user agreement is not the same as permission to use your professional style as training material for AI products. Every time an author accepts a suggestion from an AI tool, they may be contributing to a system that will eventually replace them. The line between a "tool you use" and "data that exploits you" is getting thinner.

*Meta is recognized as an extremist organization and banned in Russia.