Grammarly’s AI writing assistant used real journalists’ names without permission

Grammarly’s Expert Review feature is taking heat for using real journalists’ names without asking them first. The tool scans a draft and offers feedback supposedly shaped by named experts and trusted sources. This week, WIRED reported that Grammarly could surface dead and living writers as review personas, while The Verge found the feature using the names of real journalists and editors who never agreed to be part of it.

While Grammarly points users toward good writing or useful sources, It also puts a real person’s name next to machine-made feedback, and that changes how the advice lands. The Verge said the feature generated comments tied to Nilay Patel, David Pierce, Sean Hollister, and Tom Warren, even though none of them had agreed to be used that way.

Also read: Meta glasses face a privacy fight over claims that intimate footage was reviewed by contractors.

People will read advice differently when it comes with a familiar name attached. A plain AI suggestion is easy to skip past. A note that appears to come through the lens of a known editor or writer feels like something you should probably trust. That is where the feature gets slippery. It is not just helping with a draft. It is leaning on a reputation it did not build.

Superhuman, the company Grammarly now operates under, said the feature does not claim endorsement or direct participation and is based on publicly available, widely cited work. That may be the legal answer, but it does not clear up the product problem. In Google Docs, the suggestions reportedly looked similar to comments from real users. So the experience was not just AI giving tips. It looked a little too close to a real person stepping into your document.

The sourcing issues made that worse. The feature reportedly crashed, linked to spammy or archived pages, and in some cases pointed to sources that did not appear to match the person supposedly behind the suggestion. Some expert bios also carried outdated job titles. Due to these, the whole thing feels less like an expert review and more like a credibility costume.

The pressure from writers and academics is unlikely to die down from here, and Grammarly will probably have to tighten the product or keep wearing the blowback. For now, users should treat Expert Review the way they would treat any other AI output. Read it carefully, check the source trail, and do not mistake a famous name on the screen for real editorial judgment.