Grammarly is facing its first major legal reckoning over its controversial "Expert Review" AI feature. Journalist Julia Angwin filed a class-action lawsuit Wednesday against the writing assistant platform, alleging the company violated privacy and publicity rights by using her identity - and those of dozens of other journalists - to power AI-generated suggestions without permission. The complaint, filed against Grammarly's parent company Superhuman Platform, marks a critical test case for AI ethics in enterprise software as companies race to deploy features that blur the line between human expertise and machine-generated content.
Grammarly just learned that borrowing credibility without asking doesn't fly in court. The popular writing assistant platform is now defending a class-action lawsuit after journalist Julia Angwin discovered the company had been using her name and professional reputation to sell AI-generated writing advice to millions of users.
The complaint filed Wednesday targets Superhuman Platform Inc., Grammarly's parent company, for what Angwin's legal team describes as a brazen violation of privacy and publicity rights. The lawsuit centers on Grammarly's "Expert Review" feature, which presented AI-generated suggestions under the guise of expertise from real journalists, writers, and subject matter experts - none of whom had agreed to participate.
Angwin found out her identity was being exploited through Casey Newton's newsletter Platformer, where the journalist detailed his own discovery that Grammarly was using his name. Newton is among the experts The Verge uncovered in its investigation of the feature, which revealed dozens of professionals whose identities Grammarly appropriated without consent.
The mechanics of the scheme are straightforward but legally problematic. When Grammarly users received writing suggestions through the platform's AI tools, these recommendations appeared alongside the names and credentials of real experts - creating the false impression these individuals had personally reviewed the work. For a company that's built its reputation on helping users write with authority and credibility, the irony of manufacturing fake expert endorsements isn't lost on observers.
Wired first reported the lawsuit's filing, which comes months after The Verge's initial exposé forced Grammarly to disable the Expert Review feature. But shutting down the tool didn't erase the alleged damage to the journalists whose names and professional reputations were used to drive engagement with a commercial product.
The legal theory behind Angwin's case rests on state laws protecting individuals' right of publicity - the principle that you own your identity and can control how it's used commercially. These laws exist specifically to prevent companies from exploiting someone's name, image, or likeness for profit without compensation or consent. California, where many of the affected journalists work, has particularly strong publicity rights protections.
What makes this case especially significant for the AI industry is the precedent it could set. As companies race to differentiate their AI products, many are attaching real human credentials to machine-generated outputs. OpenAI, Google, and Microsoft have all experimented with ways to signal expertise or authority in AI responses. If Angwin prevails, it could force a reckoning across the sector about consent, attribution, and the boundaries of how AI can leverage human identity.
Grammarly's enterprise SaaS business model adds another layer of complexity. The company serves over 30 million daily users and counts major corporations among its clients. Those business customers paid for a premium product partly on the promise of expert-level writing assistance. If that expertise was fabricated through unauthorized use of real people's identities, it raises questions about what exactly customers were buying.
The lawsuit seeks class-action status, which could potentially include dozens of other journalists and experts whose names Grammarly used. Each individual affected could be entitled to damages under publicity rights laws, which can include statutory damages and profits attributable to the unauthorized use.
For Grammarly, the timing is particularly awkward. The company has been positioning itself as a responsible AI player in the enterprise market, emphasizing security, privacy, and ethical AI deployment. Those marketing messages now sit uncomfortably alongside allegations the company casually appropriated journalists' professional identities to enhance its product.
Industry watchers note this case arrives as regulatory scrutiny of AI intensifies. While most AI regulation focuses on issues like copyright infringement in training data or algorithmic bias, the Grammarly lawsuit highlights a different concern: how AI companies use real people's identities and expertise to market and differentiate their products. It's one thing to train an AI on publicly available text. It's another to explicitly attach real people's names to AI outputs without their knowledge.
The complaint doesn't specify damages sought, but publicity rights cases can result in substantial awards, especially when the unauthorized use is ongoing and commercial in nature. For a company valued in the billions - Grammarly raised funding at a $13 billion valuation in 2021 - even a significant settlement might be manageable. The reputational damage and regulatory attention, however, could prove more costly.
This lawsuit represents more than one journalist's fight to control her professional identity. It's a potential inflection point for how AI companies can ethically deploy features that trade on human expertise and credibility. As enterprise AI tools become more sophisticated, the temptation to borrow credibility from real experts will only grow. Angwin's case forces the industry to confront whether consent and compensation should be prerequisites for that borrowing, or whether AI companies can simply help themselves to whatever public identities serve their product roadmaps. The answer will shape not just Grammarly's future, but the boundaries of acceptable AI development across the sector.
