The rise and fall of Grammarly’s ‘Expert Review’ AI feature

This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on the ups and downs of AI, follow Stevie Bonifield. The Stepback arrives in our subscribers’ inboxes at 8AM ET. Opt in for The Stepback here.

Most people probably know Grammarly for its browser extension that suggests how to spruce up your emails, but over the past few years, it’s been eyeing bigger ambitions. In October, the company formerly known as Grammarly made a public pivot to rebrand as an AI company called Superhuman. The new name was adopted from Superhuman Mail, an AI email platform that Grammarly acquired in June 2025.

Superhuman CPO Noam Lovinsky vowed that “the Grammarly brand isn’t going anywhere.” Grammarly would live on as part of Superhuman, but the writing aid’s sidebar would increasingly become a hub for AI agents, rather than just grammar and spelling suggestions.

One of the rebrand’s most contentious elements actually appeared a few months prior to that big announcement. In August 2025, Grammarly quietly launched a feature called “Expert Review,” which according to a now-removed help page, offered users “insights from leading professionals, authors, and subject-matter experts.”

When a Grammarly user selected the Expert Review button, the feature would generate suggestions “inspired by” relevant experts, under their names alongside a checkmark icon. (What this verified-style icon was supposed to mean remains a mystery.) Screenshots on the feature’s help page showed it using the names of Stephen King, Neil deGrasse Tyson, and Carl Sagan, among other famous writers and academics.

The side panel for Expert Review contained a subtle disclaimer stating that references to the experts in the feature “do not indicate any affiliation with Grammarly or endorsement by those individuals or entities.”

The feature went largely unnoticed for several months, flying under the radar until March 4th, when Wired reported that it had been spotted using the names of deceased professors to give writing feedback.

In early March, a couple of us at noti.group tried out Expert Review. All it took was feeding the feature a few drafts of Verge articles before we started seeing our own colleagues’ names emblazoned on Grammarly’s AI-generated suggestions. Nilay Patel, David Pierce, Tom Warren, and Sean Hollister were spotted instantly.

None of them gave Grammarly permission to use their likenesses in its feature. On top of that, the suggestions under their names were pretty obtuse, if not annoying — for instance, headline advice inspired by “Nilay Patel” called for “urgency” and “intrigue” by suggesting generic word salad.

When noti.group asked if Superhuman thought about notifying the real people “inspiring” these Expert Reviews, Alex Gay, vice president of product and corporate marketing at Superhuman, deflected, instead saying, “The experts in Expert Review appear because their published works are publicly available and widely cited.” However, Grammarly itself seemed to have trouble citing its sources since in our testing, the “source” links on the Expert Review suggestions were often broken or redirected to completely unrelated articles.

On March 10th, a few days after noti.group reported finding our staff members’ names in Expert Review, Grammarly responded by launching an email inbox where experts could opt out of the feature. At the time, there was no indication that Superhuman was planning to disable the feature entirely or give the experts whose names they were using any modicum of control beyond firing off an email to request that their likeness not be used.

But the next day, Grammarly pivoted and announced that it would be disabling Expert Review, after all. Ailian Gan, Superhuman’s director of product management, commented on the change in a statement to noti.group, saying, “After careful consideration, we have decided to disable Expert Review as we reimagine the feature to make it more useful for users, while giving experts real control over how they want to be represented — or not represented at all.”

Superhuman CEO Shishir Mehrotra also responded in a post on LinkedIn, stating, “we received valid critical feedback from experts who are concerned that the agent misrepresented their voices.” Mehrotra went on to add, “We hear the feedback and recognize we fell short on this. I want to apologize and acknowledge that we’ll rethink our approach going forward.” Despite the apology, furious LinkedIn users continued to pile on Mehrotra’s post.

Following the demise of Expert Review, Mehrotra appeared on Decoder, where noti.group’s Nilay Patel confronted him about Grammarly’s use of his name without permission. Mehrotra stated repeatedly that Expert Review was a “bad feature,” as well as a “buried” one. (“It had very little usage.”) He also claimed that Grammarly was actually only referencing Nilay in attributions to his work.

“There’s a very thin line between taking publicly available work and being able to refer to it, and copying it,” Mehrotra said, adding, “And if you drew a line that attributing something is like using their name and likeness, then it’s a very hard line to draw.”

“This wasn’t an attribution,” Nilay replied. “You just made something up and put my name on it. There’s no attribution here. This isn’t anything I ever said. It’s not something I would ever say. I’m not even sure how you would get to the idea that based on my work that I would ever say anything like this.”

Grammarly had “source” links but, as mentioned previously, those links were often broken or, referred to content that contained no mention of editing practices or advice. Grammarly can let its users generate as many AI writing tips as they want; the issue here was using the names of Verge staffers and countless other writers, journalists, and academics to give those suggestions an appearance of authority that they didn’t really have.

The same day Superhuman announced it was shutting down Expert Review, investigative journalist Julia Angwin filed a class action lawsuit against Superhuman. The lawsuit alleged that Superhuman violated her privacy and publicity rights, as well as those of the other people named in the Expert Review feature, and broke likeness protection laws in New York and California.

Meanwhile, Expert Review appears to be gone for the time being. The feature is no longer available in Grammarly, although it sounds like it might not be offline permanently. Mehrotra’s apology post on LinkedIn seems to suggest that Superhuman hopes to “reimagine the feature” and potentially relaunch it one day: “For experts, this is a chance to build that same ubiquitous bond with users, much like Grammarly has. But in this world, experts choose to participate, shape how their knowledge is represented, and control their business model. That future excites me, and I hope to build it with experts who want to develop it alongside us.”

During his Decoder interview, Mehrotra also suggested that the future of the creator economy could be something like Expert Review, where creators (or experts) train AI agents to represent them and interact with audiences on their behalf, like by editing their writing. It’s clear AI is going to have some impact on creators, but it seems like following the Expert Review template probably isn’t going to land well with audiences.

More than anything, though, Grammarly’s Expert Review feature serves as a case in point for how people feel about generative AI at the moment. Superhuman ingested countless subject matter experts’ work then used it to generate writing suggestions with AI, put those experts’ names on those suggestions, offered the feature to paying subscribers, and didn’t get consent from the people whose names were the main draw for the feature, let alone compensate those people. It’s a clear-cut example of the extractive nature of AI.