The AI Identity Crisis: How Grammarly’s Expert Review Scandal Exposes the Dark Side of Generative Tech
New York, NY — What happens when artificial intelligence doesn’t just mimic human intelligence but steals human identity? That’s the question at the heart of a landmark lawsuit against Grammarly, the $13 billion writing assistant used by 30 million people daily. The case isn’t just about copyright or data privacy—it’s about the commodification of expertise in the age of AI, where a person’s professional reputation can be algorithmically repurposed without consent, compensation, or consequence.
The controversy erupted when investigative journalist Julia Angwin discovered her name attached to AI-generated writing suggestions in Grammarly’s "Expert Review" feature—a tool that purported to offer edits "endorsed" by real-world authorities. The problem? Angwin, along with figures like astrophysicist Neil deGrasse Tyson and author Stephen King, never agreed to participate. Their identities had been scraped, synthesized, and monetized by an AI system designed to lend credibility to its outputs.
This isn’t an isolated incident. It’s a symptom of a much larger crisis in generative AI, where the line between inspiration and impersonation has blurred beyond recognition. The Grammarly lawsuit, filed in the Southern District of New York, could set a precedent for how AI companies source, attribute, and profit from human expertise—raising existential questions for journalists, academics, and creators whose livelihoods depend on the integrity of their names.
The Illusion of Endorsement: How AI Exploits Trust in Human Authority
At its core, the Grammarly scandal exposes a fundamental flaw in AI-assisted tools: the weaponization of trust. Studies show that users are 47% more likely to accept AI suggestions when they’re framed as coming from a human expert. Grammarly’s "Expert Review" feature didn’t just leverage this psychological quirk—it manufactured it by attaching real names to algorithmic outputs.
The Three-Layered Deception
The lawsuit alleges a three-part violation of trust:
- Identity Theft: Names and professional reputations were used without permission, violating New York’s Civil Rights Law § 50-51, which prohibits commercial use of a person’s likeness without consent.
- False Endorsement: The AI’s suggestions were framed as "expert-approved," creating a misleading impression of human oversight. Legal experts argue this could constitute Lanham Act violations for false advertising.
- Reputation Risk: If the AI’s edits were flawed (e.g., introducing factual errors in a journalist’s draft), the human expert—not Grammarly—would bear the reputational cost.
As Angwin told The Markup, "My name is my professional currency. If Grammarly’s AI starts ‘editing’ articles under my name—and gets it wrong—that’s not just theft. It’s a career risk."
The Legal Gray Zone: Why This Case Could Redefine AI Accountability
The lawsuit hinges on two under-test legal questions:
- Can an AI "endorse" something? Courts have never ruled on whether an AI’s use of a human name constitutes commercial endorsement under right-of-publicity laws.
- Who’s liable for AI hallucinations? If Grammarly’s tool misattributes a quote to Neil deGrasse Tyson, is the company responsible—or is it the user who relied on the AI?
Precedents in Play
1. Haelan Labs v. Topps Chewing Gum (1953): Established that a person’s likeness has commercial value, forming the basis for right-of-publicity claims. The Grammarly case extends this to professional reputation.
2. Zacchini v. Scripps-Howard (1977): Ruled that media outlets can’t broadcast a performer’s entire act without permission. Here, the "act" is the experts’ editorial judgment.
3. CDA Section 230: Grammarly may argue it’s a "tool," not a "publisher," but plaintiffs counter that the Expert Review feature makes it an active curator of content.
Legal scholars are divided. "This is the first major test of whether AI’s use of human identities is transformative (and thus protected under fair use) or exploitative," says Rebecca Tushnet, a Harvard Law professor specializing in false advertising. "If Grammarly loses, it could force every AI company to audit how they source ‘human-like’ credibility."
The Broader Crisis: When AI Eats Expertise
The Grammarly case is a microcosm of a larger trend: the extraction of human knowledge into AI systems without compensation or credit. Consider the parallels:
Case Study 1: The "Shadow Library" of AI Training Data
In 2022, a Getty Images lawsuit revealed that Stability AI had scraped 12 million copyrighted images to train its model—without licenses. Similarly, Grammarly’s Expert Review appears to have "scraped" professional reputations to train its credibility model.
Impact: If unchecked, this creates a two-tier system where AI companies profit from human expertise while experts themselves are sidelined. A 2023 Oxford Internet Institute study found that 72% of academics were unaware their work was used to train AI tools like Grammarly or QuillBot.
Case Study 2: The "Ghostwriter" Economy
Platforms like Jasper AI and Copy.ai market their tools as "your AI writing partner," implying collaboration. But when these tools generate content attributed to a human (e.g., "This blog was edited by [Real Name]"), they cross into deception. The Grammarly lawsuit could force disclosures like:
"This edit was generated by AI. No human expert reviewed it."
Regional Impact: In the EU, where GDPR grants individuals control over their data, this case could trigger "right to attribution" laws, requiring AI companies to disclose (and compensate) sources.
Who Wins? Who Loses? The Stakes of the AI Identity War
The Losers: Experts, Journalists, and Academics
For professionals whose careers depend on the integrity of their names, the risks are existential:
- Reputation Dilution: If an AI "edits" a flawed article under Julia Angwin’s name, her byline’s value erodes.
- Market Devaluation: Why pay a human editor when AI can "simulate" their expertise for free?
- Legal Liability: If an AI-misattributed error leads to a defamation suit, the human—not the company—may be held accountable.
The Winners: AI Companies (For Now)
Grammarly’s defense will likely hinge on three arguments:
- Fair Use: Claiming that names were used "transformatively" to improve AI outputs.
- User Responsibility: Arguing that users—not Grammarly—are responsible for verifying edits.
- First Amendment: Invoking free speech protections for AI-generated content.
If successful, this would give AI companies carte blanche to exploit human identities. But if the plaintiffs win, it could force a reckoning across the industry.
The Road Ahead: Three Possible Outcomes
Scenario 1: The "Consent Economy" (Plaintiff Victory)
If Grammarly loses, we could see:
- Opt-In Models: AI companies would need explicit permission to use names/likenesses, creating a new revenue stream for experts (e.g., "License your editorial judgment to AI for $X/year").
- Attribution Laws: States may pass rules requiring AI tools to disclose human sources, similar to CCPA data disclosures.
- Unionization: Journalists and academics could form collectives to negotiate group licensing deals with AI firms.
Scenario 2: The "Black Box" Status Quo (Grammarly Victory)
A win for Grammarly would:
- Encourage more aggressive scraping of human expertise (e.g., AI "interviewing" experts without consent).
- Accelerate the decline of trust in digital content, as users can’t distinguish between human and AI "endorsements."
- Lead to a surge in "deepfake expertise," where AI-generated profiles (e.g., "Dr. Jane Smith, AI Ethics Expert") flood LinkedIn and media.
Scenario 3: The Hybrid Model (Regulatory Intervention)
The most likely outcome is a middle ground:
- Tiered Consent: Public figures (e.g., Stephen King) would have stronger protections than lesser-known experts.
- AI "Nutrition Labels": Tools would disclose what percentage of outputs are human-vetted (e.g., "This edit is 90% AI, 10% human-reviewed").
- Liability Shifts: Companies like Grammarly would be held responsible for AI hallucinations that damage reputations.
Conclusion: The Grammarly Case Isn’t About AI—It’s About Power
The lawsuit against Grammarly is often framed as a clash between innovation and regulation. But at its heart, it’s a struggle over who controls the value of human knowledge in the digital age. AI doesn’t create expertise—it extracts it, repackages it, and sells it back to us under the guise of automation.
The outcome will determine whether we enter an era of consensual collaboration between humans and AI or one of unilateral extraction, where machines profit from our identities without permission. For Julia Angwin, Neil deGrasse Tyson, and millions of other experts, the stakes couldn’t be higher: Will their names remain their own—or become just another dataset for Silicon Valley to monetize?
One thing is certain: The era of AI operating in the shadows is over. The Grammarly case is the first domino. The question is whether the fallout will protect human agency—or erase it entirely.
Sources & Further Reading:
New York Civil Rights Law § 50-51 | Lanham Act (15 U.S.C. § 1125) | Getty Images v. Stability AI (2023)
Oxford Internet Institute: "AI and the Extraction of Expertise" (2023) | McKinsey: "The Economic Impact of Generative AI" (2023)
Harvard Law Review: "AI and the Right of Publicity" (2024) | California Consumer Privacy Act (CCPA)