You probably know Grammarly as a simple browser extension that helps clean up your emails and correct grammatical errors, but over the past few years, the company has set its sights on much bigger ambitions in the tech world. In October, the company formerly known as Grammarly took the bold public step of rebranding as an artificial intelligence company named Superhuman. The new name was adapted from Superhuman Mail, an AI-powered email platform acquired by the company in June 2025.
The company’s chief product officer, Naom Lovinsky, pledged that the original brand would not disappear and would instead continue as part of the new entity, with the writing assistance sidebar gradually transforming into an all-in-one hub for AI agents rather than just offering spelling and grammar suggestions.
How did the crisis begin behind the scenes?
One of the most controversial elements of this rebrand appeared a few months before the major public announcement. In August 2025, Grammarly quietly rolled out a feature called “Expert Review.” According to a help page that was later deleted, the feature provided users with “insights from top professionals, authors, and subject-matter experts.”
When a user clicked the review button, the tool generated suggestions and edits supposedly “inspired” by real experts, listing them under their names alongside a suspicious verification symbol (the true meaning of which remains a mystery today). Promotional screenshots for the feature showed the names of prominent figures such as Stephen King, Neil deGrasse Tyson, and Carl Sagan, alongside a group of famous writers and academics.
The feature’s sidebar contained a veiled disclaimer stating that mentioning these experts “does not imply any affiliation with or endorsement by those individuals or entities.” The feature went largely unnoticed for months until March 4, when Wired published a report revealing the tool was caught using the names of deceased university professors to provide writing feedback.
Uncovering the trick and journalists’ fury
In early March, a team of journalists at The Verge tested the feature themselves. It took only entering a few article drafts before they began seeing their colleagues’ names popping up directly on AI-generated suggestions. Prominent names like Nilay Patel, David Pierce, Tom Warren, and Sean Hollister were spotted immediately.
The major issue was that none of these journalists had granted the company permission to use their identities. Furthermore, the suggestions listed under their names were superficial and annoying; for instance, headline-writing advice “inspired by Nilay Patel” demanded a sense of “urgency” and “excitement,” suggesting a bunch of random, useless words.
When asked whether the company considered notifying the real people who “inspired” these reviews, Alex Gay, vice president of product marketing, sidestepped a direct answer, stating: “Experts appear in the feature because their published work is publicly available and widely cited.” However, the tool itself seemed to struggle with citing its sources, as source links attached to suggestions were frequently broken or directed users to unrelated articles.
Administrative blunders and delayed apologies
On March 10, just days after the exposure, management responded by launching an email address where experts could request to be excluded from the feature. At the time, there was no indication of any intention to disable the feature entirely or give experts any real control over their identity other than submitting an opt-out request.
However, just a day later, the company reversed course and announced it would disable the feature. Aileen Gan, director of product management, commented: “After careful consideration, we have decided to disable the Expert Review feature while we reimagine it to be more useful, while giving experts genuine control over how they are represented, or not represented at all.”
CEO Shishir Mehrotra also published a statement on LinkedIn, saying: “We received valid critical feedback from experts who fear the AI agent may have distorted their voices. We are listening to that feedback and recognize we fell short here. I want to apologize and affirm that we will rethink our approach going forward.” Despite this apology, angry users continued to heavily criticize the post.
Direct confrontation and legal repercussions
Following the shutdown of the feature, Mehrotra appeared on Decoder, where Nilay Patel confronted him regarding the unauthorized use of his name. Mehrotra reiterated that the feature was a “bad idea” and “buried,” having seen little use, and claimed the tool was merely referencing Nilay’s work.
But Patel fired back firmly: “This wasn’t attribution. You manufactured something and put my name on it. This is not something I ever said, or something I would ever say.” This confrontation highlighted the exploitative core of the feature, which used the names of journalists and academics to lend AI suggestions an illusion of authority and credibility.
What does the future hold?
On the same day the shutdown was announced, investigative journalist Julia Angwin filed a class-action lawsuit against the company, alleging violations of privacy and publicity rights, as well as breaches of New York and California personal identity protection laws.
Although the feature is currently gone, Mehrotra’s statements indicate the company hopes to relaunch it someday with the voluntary participation of experts. This crisis serves as a vivid example of the ethical challenges of generative AI, and how collecting and using data without consent can damage company reputation and undermine user trust.
- On the sidelines: It turned out the feature’s sources were bypassing paywalls to read protected articles.
- Before taking his current role, Mehrotra served as the CEO of Coda, which was acquired in December 2024.
- The term “stochastic cloning” emerged in tech circles to describe this crisis, expressing the inaccurate replication of writers’ identities using AI.
Frequently Asked Questions
What is the “Expert Review” feature launched by the company?
It is an AI-powered tool designed to provide users with writing advice and edits, falsely attributing this advice to prominent figures and well-known writers to lend it an air of credibility.
Why did this feature anger writers and journalists?
Because the company used these experts’ names and identities without obtaining their permission or consent, generating superficial advice that did not reflect their actual writing style, which they viewed as a violation of their rights and an exploitation of their reputation.
Was the feature permanently shut down?
The company announced a temporary shutdown of the feature in response to widespread criticism and lawsuits, but management hinted at the possibility of redesigning and launching it in the future in a way that ensures greater control and voluntary participation for experts.
Investigative journalist Julia Angwin filed a class-action lawsuit against the company, accusing it of violating privacy and publicity rights and breaching personal identity protection laws in New York and California.
