The Tale of Grammarly’s Doppelgänger
This is The Stepback, a weekly newsletter diving into a key story from the tech landscape. For insights on the evolving world of AI, keep an eye on developments from industry experts. The Stepback lands in subscribers’ inboxes at 8 AM ET.
**How it started**
Grammarly is widely recognized for its browser extension that enhances email writing, but in recent years, it has set its sights on broader goals. In October, the company rebranded itself as Superhuman, shifting its focus to AI. This new identity stems from the acquisition of Superhuman Mail, an AI email platform, in June 2025.
Superhuman’s Chief Product Officer, Noam Lovinsky, assured users that the Grammarly brand would continue to exist within the Superhuman ecosystem. The familiar writing assistant would evolve, transforming its sidebar into a hub for AI agents rather than just a tool for grammar and spelling corrections.
One of the most controversial aspects of this rebranding emerged a few months earlier. In August 2025, Grammarly quietly introduced a feature called “Expert Review,” which claimed to provide users with “insights from leading professionals, authors, and subject-matter experts,” according to a now-removed help page.
When users clicked the Expert Review button, the feature generated suggestions “inspired by” various experts, displaying their names alongside a checkmark icon. The meaning of this verification-style icon was unclear. Screenshots from the help page showcased names like Stephen King, Neil deGrasse Tyson, and Carl Sagan, among others.
The Expert Review sidebar included a discreet disclaimer stating that references to these experts did not imply any affiliation with Grammarly or endorsement from those individuals.
Initially, the feature went largely unnoticed until March 4th, when it was reported that it had been using the names of deceased academics to provide writing feedback.
**How it’s going**
In early March, a few team members tested Expert Review. After submitting drafts of articles, they quickly noticed their colleagues’ names appearing in Grammarly’s AI-generated suggestions. Names like Nilay Patel, David Pierce, Tom Warren, and Sean Hollister were prominently featured.
None of these individuals had given permission for their names to be used. Moreover, the suggestions attributed to them were often vague or frustrating—one suggestion inspired by “Nilay Patel” urged for “urgency” and “intrigue” but offered generic advice instead.
When asked if Superhuman considered notifying the individuals whose names were used in Expert Review, Alex Gay, the vice president of product and corporate marketing at Superhuman, deflected, stating, “The experts in Expert Review appear because their published works are publicly available and widely cited.” However, during testing, the “source” links provided with the suggestions were frequently broken or led to unrelated content.
On March 10th, shortly after the names of staff members were reported in Expert Review, Grammarly launched an email inbox for experts to opt out of the feature. At that time, there was no indication that Superhuman intended to disable the feature entirely or provide experts with any real control beyond sending an email to request their names be removed.
However, the following day, Grammarly announced it would indeed disable Expert Review. Ailian Gan, Superhuman’s director of product management, explained that the decision came after careful consideration, aiming to reimagine the feature to better serve users while giving experts genuine control over their representation.
Superhuman CEO Shishir Mehrotra also addressed the situation, acknowledging the valid concerns raised by experts about misrepresentation. He expressed a commitment to reevaluating their approach moving forward. Despite the apology, many users continued to voice their frustrations on LinkedIn.
After the discontinuation of Expert Review, Mehrotra appeared on a podcast where he was confronted about the unauthorized use of names. He described Expert Review as a “bad feature” and claimed it had minimal usage. He insisted that Grammarly was merely referencing the work of individuals rather than misusing their names.
“There’s a very thin line between taking publicly available work and being able to refer to it, and copying it,” Mehrotra stated, adding that if one equated attribution with using someone’s name and likeness, it created a challenging distinction.
“This wasn’t an attribution,” Nilay countered. “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.”
Grammarly’s “source” links were often broken or led to content that didn’t provide any editing advice. While the platform allows users to generate AI writing tips, the core issue was using the names of various writers, journalists, and academics to lend authority to suggestions that lacked credibility.
**What happens next**
On the same day Superhuman announced the shutdown of Expert Review, investigative journalist Julia Angwin filed a class action lawsuit against the company. The lawsuit claimed that Superhuman violated privacy and publicity rights, as well as likeness protection laws in New York and California.
For now, Expert Review seems to be offline, but it may not be gone for good. Mehrotra’s apology hinted at a potential reimagining and relaunch of the feature in the future, emphasizing a desire for experts to have a say in how their knowledge is represented.
During his podcast interview, Mehrotra suggested that the future of the creator economy could resemble Expert Review, where creators train AI agents to represent them and engage with audiences. While AI is poised to impact creators, it’s clear that following the Expert Review model may not resonate well with users.
Ultimately, Grammarly’s Expert Review feature highlights the current sentiment surrounding generative AI. Superhuman utilized the work of numerous experts to generate writing suggestions, attaching their names without consent or compensation, exemplifying the extractive nature of AI.
