Grammarly borrows journalists' identities for AI writing tips

Grammarly just learned the hard way that slapping famous journalists' names on AI suggestions without asking isn't "expert advice"—it's a PR nightmare.

Back in August 2025, Grammarly rolled out an "Expert Review" feature inside its writing assistant sidebar. The tool claimed to sharpen users' drafts "through the lens of industry-relevant perspectives." But when tech reporters actually opened it up, they found AI-generated comments presented as if real people—including their own editors, living and dead authors, and renowned journalists—were marking up their Google Docs.

Fake edits from real people

None of the featured people had agreed to be in the tool. The roster included big names like Stephen King, Carl Sagan, Kara Swisher, Bloomberg's Mark Gurman, and The Verge's editor-in-chief Nilay Patel.

In Google Docs, Grammarly went so far as to format these AI-generated tips to mimic real document comments, complete with check-mark logos. The advice itself often completely missed the mark. One Verge writer noted that an AI edit supposedly inspired by senior editor Sean Hollister advised adding redundant parenthetical context—something the real Sean routinely cuts during actual edits.

Beyond that, the feature's job titles were frequently outdated, the software crashed, and the underlying "source" links often routed users to broken archives or outright spam sites that had nothing to do with the cited authors.

Superhuman, Grammarly's parent company, quickly went on defense. Alex Gay, vice president of product and corporate marketing, told The Verge that the experts were selected simply because their published work is "publicly available and widely cited," adding that the tool never claimed direct endorsements. Grammarly's fine print also insists these mentions are merely for informational purposes.

As historian C.E. Aubin blunt told Wired, that disclaimer falls flat: "These are not expert reviews, because there are no 'experts' involved in producing them."

Why it matters

We think Grammarly crossed a blatant boundary here, substituting actual product depth with hijacked credibility.

For regular writers and professionals, this feature offers zero real value right now. It's essentially generic prompt engineering dressed up with a famous byline. Relying on it won't give you Nilay Patel's critical edge; it just gives you boilerplate AI text that pretends to sound like him.

For the wider tech industry, this is an object lesson in what happens when generative AI interfaces lean into deceptive design. It is one thing to train a model on public writing, but quite another to simulate a specific living person's active feedback in a collaborative work environment. By turning respected names into unlicensed avatars, Grammarly traded its reputation as a trusted utility for a gimmicky feature that will almost certainly invite renewed legal and regulatory scrutiny over AI impersonation.