A class action lawsuit filed at the U.S. District Court for the Southern District of New York in 2026 is revealing new legal issues in the generative AI era. Investigative journalist Julia Angwin filed a class action against Superhuman Platform, which operates AI writing service Grammarly, claiming that its AI features used the names and identities of hundreds of writers and journalists without authorization.
The core of this lawsuit is the "Expert Review" feature Grammarly introduced in 2025. This service was marketed as providing feedback based on the style and expertise of famous authors or journalists when users upload writing. The problem is that the names of actual individuals were displayed during the feedback process.
According to the plaintiff's side, individuals displayed as feedback providers included not just Julia Angwin but also famous authors and journalists such as novelist Stephen King. However, many of them had not consented to their names or identities being used in the feature.
The plaintiff's side argued in the complaint that "Grammarly commercially exploited the names and identities of writers, and AI expressed non-existent advice as if those individuals had provided it." This case is evaluated as a representative example that simultaneously reveals issues of knowledge, identity, and copyright in the generative AI era beyond a simple service dispute.
Particularly noteworthy in this case is that the legal basis of the lawsuit is the Right of Publicity rather than traditional copyright. The Right of Publicity is the right that requires obtaining the consent of the individual concerned when commercially using a person's name, image, voice, identity, etc. U.S. courts have long judged the act of using another person's name for advertising or commercial purposes as illegal, and in particular, commercial use without consent is often considered privacy invasion.
New York Civil Rights Law Section 50 also clearly stipulates this principle. The provision designates the act of using an individual's name or portrait for advertising or commercial purposes without consent as illegal. Ultimately, the core question of this case is compressed into the problem of "can AI imitate someone's voice or expertise?"
The development of generative AI is shaking the existing intellectual property protection system. The existing creative work protection system is broadly divided into three categories. Copyright protects creative works, trademarks protect brands, and patents protect technology. However, generative AI is creating new problems that transcend these distinctions.
AI has the characteristic not of simply copying existing text but of learning a specific author's style, tone, knowledge structure, and even identity. For example, it has become possible to generate writing in the style of a specific author, provide advice like a specific expert, or mimic the perspective of a specific journalist. This change creates a new legal question: "How far can AI learning someone's knowledge and style and speaking like that person be permitted?"
AI companies have generally presented similar logic regarding such issues. Companies argue that AI does not copy specific individuals but merely learns statistical patterns from vast data. They also explain that AI-generated outputs are new content, not verbatim reproductions of existing data. This logic has also appeared in various AI copyright lawsuits.
Representative examples include The New York Times' copyright lawsuit against OpenAI, the Authors Guild's lawsuits against various AI companies, and Getty Images' image copyright lawsuit against Stability AI.
However, the Grammarly case shows a different direction from existing copyright disputes. Here, the center of the debate is the use of identity, not data training itself. That is, there is the argument that AI learning a specific writer's style and AI expressing things "as if this writer advised it" are completely different problems.
Experts see this case as potentially presenting a new direction for AI regulation. AI regulation debates until now have mainly focused on copyright issues of training data, the legality of data scraping, and fair use issues. However, analysis is emerging that new legal concepts such as style rights, identity rights, and expertise rights are likely to appear going forward.
Style rights relate to the problem of AI imitating a specific creator's expression method or writing style. Identity rights refer to the problem of AI speaking or advising like a specific individual. Also, expertise rights are connected to the problem of AI utilizing a specific expert's authority to secure credibility. The Grammarly lawsuit is evaluated as an important case in that it simultaneously raises all three problems.
The verdict in this case could also affect the AI industry as a whole. Not just AI writing platforms but various industry areas including AI content generation services and AI agent services providing expert advice could face the same problem. In particular, as the AI Agent era fully arrives, AI will develop from a simple text generation tool into a system that advises and analyzes like experts.
At this point, the core question that arises is "in whose name does AI speak?" Going forward, if AI comes to make diagnoses like doctors, analyze like lawyers, and explain like journalists, who the expertise identity and responsibility belong to will likely become an increasingly important debate.
Analysis is also emerging that current AI-related legal disputes are developing in three stages. The first stage is training data copyright disputes. The second stage relates to the boundary between model output and original works. And the third stage, now beginning, is the dispute over identity rights and expertise rights. The Grammarly case is at the forefront of the third stage. What kind of precedent the courts will set in this process will be an important signal for the direction of AI regulation going forward.
