Grammarly pulls AI author-impersonation tool after backlash
Writing tool Grammarly has disabled an AI feature which mimicked personas of prominent writers, including Stephen King and scientist Carl Sagan, following a backlash from people impersonated.
The Expert Review function, which offered writing feedback "inspired by" the styles of famous authors and academics, was taken down this week by Superhuman, the tech firm which runs Grammarly.
The feature was met with resistance, including a multi-million dollar lawsuit, from writers who found their names and reputations used as "AI personas" without their consent.
Shishir Mehrotra, the firm's chief executive, apologised on LinkedIn, acknowledging the tool had "misrepresented" the voices of experts.
Investigative journalist Julia Angwin, a New York Times contributing opinion writer, is the lead plaintiff in a class-action lawsuit filed against Superhuman and Grammarly in the Southern District of New York.
Angwin told the BBC she was "stunned" to find her professional identity being marketed as a commercial product.
"I had thought of deepfakes as something that happens to celebrities, mostly around images," she said.
"Editing is a skill... it's my livelihood, but it's not something I've ever thought about anyone trying to steal from me before. I didn't even think it was stealable."
The legal filing alleges that the tech firm misappropriated the identities of "hundreds" of writers to drive profits for its paid subscription service.
According to Angwin's lawyer, Peter Romer-Friedman, the case has already seen significant momentum.
"We've heard from over 40 people in the last 24 hours since we filed the suit," he said, describing the company's actions as a "brazen violation of the law."
The lawsuit argues it is unlawful to use names for commercial purposes without consent and seeks to stop the platform from attributing advice to experts that they "never gave."
'Slopperganger'
While the filing states that damages exceed $5m (£3.7m), Romer-Friedman noted that this is a minimum jurisdictional requirement and the true figure will be calculated based on the firm's earnings from the tool.
For Angwin, the quality of the AI's output added insult to injury, describing the imitation as a "slopperganger" - a reference to content described on social media as "AI slop".
"The edits were not good. The ones that they were attributing to me... were making the sentences worse, more complex," she said.
"The idea that my name would be in there giving people terrible advice is actually really appalling."
Grammarly was founded in 2009 as a writing-review tool and began integrating a suite of generative-AI tools in August 2025.
Part of this was the Expert Review function which appears to have launched without the named famous personas introduced later.
Although the company began rebranding to Superhuman in October, Grammarly was kept as the name of its main service.
As criticism mounted in recent days, Superhuman initially said it would maintain the feature but allow those named to "opt-out", according to The Verge.
Wes Fenlon, a gaming journalist whose persona was used in the tool, wrote on BlueSky: "Opt-out via email is a laughably inadequate recourse for selling a product that verges on impersonation and profits on unearned credibility."
Romer-Friedman argued that the burden of consent should never have been on the writers.
'We fell short'
Mehrotra said in response to the backlash: "Over the past week, we received valid critical feedback from experts who are concerned that the agent misrepresented their voices.
"This kind of scrutiny improves our products, and we take it seriously."
He said the AI agent had drawn on "publicly available information from third-party LLMs to surface writing suggestions inspired by the published work of influential voices".
The firm's chief executive apologised, adding: "We hear the feedback and recognize we fell short on this."
Responding to the lawsuit, Mehrotra told the BBC: "We announced that Expert Review was being taken down for a redesign before the claim was filed, and in its short lifespan it had very little usage.
"We are sorry, and we will rethink our approach going forward."
However, he said the legal claims within the lawsuit are "without merit" and the company will "strongly defend against them".
He added the firm is working on a "better approach to bringing experts onto our platform" in a way that will "benefit both users and experts".
Sign up for our Tech Decoded newsletter to follow the world's top tech stories and trends. Outside the UK? Sign up here.
Facts Only
* Superhuman disabled its “Expert Review” AI feature.
* The feature mimicked the writing styles of Stephen King and Carl Sagan.
* The lawsuit was filed by Julia Angwin, a New York Times contributor.
* The lawsuit alleges misappropriation of writers’ identities for profit.
* Superhuman CEO Shishir Mehrotra apologized for the misrepresentation.
* The legal claim includes a minimum jurisdictional demand of $5 million.
* Grammarly was founded in 2009 and began integrating generative AI in August 2025.
* The company rebranded to Superhuman in October 2025.
* The “Expert Review” feature was taken down for redesign.
* Over 40 people have contacted the legal team since the lawsuit was filed.
* The company used publicly available information from LLMs.
* The lawsuit seeks to stop the platform from attributing advice to experts without consent.
Executive Summary
Full Take
The Grammarly situation presents a fascinating, if uncomfortable, convergence of technological hubris and the evolving nature of authorship. The core problem isn't simply the imitation of styles—the fundamental flaw lies in the *unauthorized* application of individual intellectual property as a commercial product. This echoes patterns we've seen before with deepfakes, but expands the scope significantly, targeting not just visual representation, but the very *voice* of an established thinker. The “slopperganger” description from Angwin brilliantly captures the damage – not just a poorly written imitation, but a degradation of her professional identity. This isn’t simply a bad AI; it’s a fundamentally unethical extraction of value.
The legal action, spearheaded by Angwin, isn’t just about money; it’s a crucial test case for defining ownership in the age of generative AI. The fact that the initial attempt at a solution—allowing “opt-out”—was deemed “laughably inadequate” highlights a systemic failure in anticipating the potential harms. Mehrotra's apology, while acknowledging the misrepresentation, still relies on the somewhat deflective argument that the AI was drawing on "publicly available information." That’s a semantic sleight of hand – access to public information doesn’t automatically grant the right to commercialize it as a stylistic mimicry. The patterns at play here mirror the broader anxieties surrounding the use of LLMs – the potential for these systems to erode the boundaries between genuine human creativity and algorithmic simulation. The real question is not *if* this will happen again, but *how* we establish ethical guardrails before these systems further destabilize the value of intellectual labor.
Patterns detected: ARC-0024 Ambiguity (the definition of “inspired by” is deliberately vague and creates space for exploitation), ARC-0043 Motte-and-Bailey (the company initially offered “opt-out” as a defensive tactic, then immediately dismissed the claims), ARC-0017 Misdirection (shifting focus to the AI’s “learning” process to avoid culpability).
Sentinel — Likely Human
This article presents a factual account of Grammarly’s controversy surrounding its AI ‘Expert Review’ feature, focusing on legal action and initial responses from the company. While the text leans towards a balanced presentation, stylistic elements and the nature of the cited claims suggest a predominantly human origin.”