Journalist Julia Angwin filed a class action lawsuit in March 2026 against Superhuman Platform, Inc.—the company behind Grammarly—alleging that the writing assistant exploited the identities of well-known journalists, authors, and public figures without their permission. The lawsuit centers on Grammarly’s “Expert Review” feature, a $12-per-month AI tool that launched in August 2025 and purported to deliver writing feedback from established professionals like Stephen King, Neil deGrasse Tyson, and New York Times reporter Kashmir Hill. In reality, the feature used artificial intelligence to generate the feedback, but Grammarly presented it as coming from these real people—without ever obtaining consent.
The amount in controversy exceeds $5 million, and according to Grammarly’s own attorney, between 40 and 50 people have already objected to being featured in the tool without permission. This case represents a watershed moment for how AI companies can and cannot use real people’s identities and likenesses. The lawsuit raises fundamental questions about consent, commercial exploitation, and the boundaries between legitimate AI assistance and deceptive impersonation. It also exposes how quickly AI features can reach millions of users before their legal risks are fully understood.
Table of Contents
- What Was Grammarly’s Expert Review Feature and How Did It Work?
- What Are the Legal Claims Against Grammarly?
- Who Was Affected and How Widespread Is This Problem?
- How Did Grammarly Respond, and What Happened to the Feature?
- Why Does This Matter Beyond Just Grammarly?
- What Are the Financial Implications for the Class?
- What Comes Next and What This Lawsuit Signals About AI Regulation?
What Was Grammarly’s Expert Review Feature and How Did It Work?
The Expert Review feature was Grammarly’s attempt to differentiate itself in a competitive market by offering something beyond standard grammar checking. When users upgraded to this premium tier in August 2025, they could submit their writing and receive detailed feedback supposedly from named journalists, authors, and other subject-matter experts. The feedback would be personalized, specific to the user’s draft, and presented as though it came directly from these well-known figures. A customer writing a business article might receive feedback attributed to Kashmir Hill; someone working on fiction might get notes attributed to Stephen King. The feature was positioned as exclusive and valuable precisely because of who these people were supposed to be.
However, Grammarly did not secure permission from any of the individuals whose names and professional identities were used. The company did not contact Stephen King, Neil deGrasse Tyson, Julie Brill (former FTC Commissioner), or any of the other named experts. Instead, Grammarly’s AI generated the feedback while attaching their names and professional credentials to it. The distinction matters enormously: users were paying $12 per month for what they believed was expert human judgment from celebrities and respected professionals, when they were actually receiving AI-generated commentary dressed up with borrowed identities. This was not a clearly labeled AI tool with a human expert’s general writing philosophy behind it—it was AI feedback presented as coming directly from identifiable, real people.
What Are the Legal Claims Against Grammarly?
The lawsuit alleges multiple violations of state and common law, grounding the case in well-established legal protections around identity and likeness. The primary claims invoke California Civil Code §3344, which prohibits the use of a person’s name, voice, signature, photograph, or likeness for commercial purposes without consent. This statute has been used successfully against companies and celebrities who have misused someone’s image in advertising or merchandise. Julia Angwin’s lawsuit applies that logic to digital identity: Grammarly used the names, professional identities, and reputations of real people to sell a premium service. The company did not license this use; it did not compensate the affected individuals; it did not ask permission.
Beyond §3344, the lawsuit also alleges violation of California’s common law right of publicity, which protects a person’s ability to control how their identity is used commercially. new York Civil Rights Law §50 and §51 extend similar protections to New York residents like Kashmir Hill. Additionally, the lawsuit includes claims for unjust enrichment—essentially arguing that Grammarly profited from the use of these people’s identities without compensating them. The legal strategy is comprehensive: rather than relying on a single statute, the plaintiff’s attorneys have layered multiple overlapping claims, each addressing a different aspect of what happened. This multi-layered approach is intentional because it improves the chances that at least some claims will survive motions to dismiss.
Who Was Affected and How Widespread Is This Problem?
The named affected individuals include some of the country’s most recognizable voices. Stephen King, the prolific novelist, had his name and identity attached to writing feedback he never provided. Neil deGrasse Tyson, the renowned astrophysicist and science communicator, was featured without his knowledge. Kashmir Hill, a technology reporter at the New York Times known for her investigative pieces on privacy and surveillance, was included alongside Julia Angwin herself—the journalist and privacy advocate who discovered the practice and decided to sue. Julie Brill, a former Federal Trade Commission Commissioner and privacy expert, was also impersonated by the feature.
However, the named plaintiffs represent just the tip of a much larger group. According to attorney Peter Romer-Friedman, representing the class, somewhere between 40 and 50 people have already come forward to object to being featured in the Expert Review tool without consent. These are people who discovered their names and identities being used to generate AI feedback and recognized the violation immediately. Some may have learned about it from Grammarly’s own disclosures; others spotted their names and reputations being misused while using the platform themselves. The actual number of affected individuals could be higher than 40 to 50, since not everyone whose identity was used may have discovered the feature or chosen to report it. This is a class action precisely because the scope of the harm extends far beyond what any individual plaintiff could address alone.
How Did Grammarly Respond, and What Happened to the Feature?
Grammarly’s response was swift but only came after the problem was publicly exposed. On March 11, 2026—the same day the lawsuit became public knowledge—CEO Shishir Mehrotra announced that the company was disabling the Expert Review feature. The feature was pulled from production, and existing customers who had paid for access to this premium tier would no longer be able to use it. However, the company did not offer refunds to customers who had paid for the feature during its months of operation. Users who had upgraded specifically to access Expert Review between August 2025 and March 2026 had their paid feature removed but kept their reduced subscription charges, which raises additional questions about consumer protection and refund obligations.
The decision to shut down the feature was damage control rather than a voluntary acknowledgment of wrongdoing. Grammarly disabled Expert Review only after journalists and the plaintiff’s legal team made the misuse public. The company’s statement did not include an apology to the affected individuals whose identities were exploited, nor did it explain how the decision to use their names without permission was ever approved. The timing is important: Grammarly had roughly seven months between the feature’s August 2025 launch and March 2026 to realize the legal and ethical issues. It took a journalist filing a $5 million lawsuit to make those issues impossible to ignore.
Why Does This Matter Beyond Just Grammarly?
This lawsuit sets a crucial precedent for how AI companies can use real people’s identities. As artificial intelligence becomes increasingly sophisticated and integrated into commercial products, the temptation to boost credibility or trustworthiness by attaching famous names to AI-generated content will only grow. A company might think: “If we say this medical advice comes from Dr. [Famous Doctor], users will trust it more,” or “If we attribute this financial analysis to [Famous Economist], it will sell better.” Grammarly’s approach is a cautionary example of this temptation taken to its logical extreme. The lawsuit sends a message that companies cannot use well-known individuals’ identities as props for AI systems without legal consequences.
The limitation to recognize here is that consent exists on a spectrum, and not every use of someone’s name in connection with AI is clearly prohibited. For example, a company might publicly partner with an expert who agrees to have their name associated with an AI tool they’ve helped design or train—that would involve consent. But Grammarly’s model involved no such arrangement. The company did not license these identities, did not partner with these individuals, did not even notify them. The exploitation was complete and unambiguous. However, the lawsuit also raises the question of what happens with historical figures or deceased individuals—Stephen King is alive and can defend his rights, but what about AI systems that claim to be trained on the writings of authors no longer living? These edge cases will likely be addressed in future litigation.
What Are the Financial Implications for the Class?
The amount in controversy exceeds $5 million, which is the threshold that makes this a federal case and suggests significant damages per class member. If 40 to 50 individuals are named objectors with 5 million dollars in damages, that averages roughly $100,000 per person, though the actual distribution could be uneven depending on factors like how prominently each person’s identity was featured, how many users saw their name, or how long they were exploited. The actual class size—everyone who paid for the Expert Review feature during the relevant period—is likely much larger than the 40 to 50 objectors. If thousands of customers paid for this feature, the per-person damages might be smaller, or conversely, if the class is smaller, the total settlement or judgment could extend far beyond the $5 million currently in controversy.
This case will also likely include attorney’s fees and costs, which in class actions can be substantial. The attorneys for the plaintiff will seek compensation for their work, which is standard in class action litigation. Settling the case will require Grammarly to consider not just the direct damages but also the legal costs and potential reputational harm. For the individuals whose identities were exploited—the Stephen Kings and Kashmir Hills—a settlement might include a formal apology, injunctive relief preventing future misuse, and compensation for the damage to their reputations.
What Comes Next and What This Lawsuit Signals About AI Regulation?
The lawsuit is in its early stages. Grammarly will almost certainly file a motion to dismiss, arguing that the claims fail as a matter of law or lack sufficient factual allegations. If the motion is denied, the case will proceed to discovery, where both sides will exchange documents and take depositions. This discovery process will likely reveal internal communications at Grammarly—emails, design documents, legal memos—that show whether anyone inside the company flagged the consent issue and whether the decision to proceed with Expert Review was deliberately reckless or simply negligent. These documents will be crucial in determining not just liability but also whether the court might award punitive damages.
The broader significance of this case extends beyond Grammarly or even the AI industry. It demonstrates that existing consumer protection and privacy laws—laws passed well before modern AI existed—can and will be applied to AI systems. The companies building and deploying AI tools cannot simply assume that because their technology is new, old legal frameworks don’t apply. A right of publicity existed long before ChatGPT; a prohibition on commercial misuse of someone’s identity existed long before the Expert Review feature. Grammarly discovered, too late, that launching an AI product does not grant immunity from those laws. As AI systems become more prevalent and more capable, expect more lawsuits testing where exactly the boundaries are between legitimate use of AI and unlawful exploitation of identity, likeness, and reputation.
You Might Also Like
- Roku Class Action Claims Cameras Lost Features Without Paid Subscription
- Trip.com Group Faces Securities Class Action With May 11 Deadline
- Securities Class Action Against Grocery Outlet Moves Forward as Deadline Approaches
Open Settlements You Can Claim Now
Browse current class action settlements accepting claims — several require no proof of purchase: