Grammarly’s AI Uses Dead Authors’ Styles Without Permission in 2026

Grammarly’s AI Uses Dead Authors’ Styles Without Permission

Grammarly, now operating under its newly rebranded parent company Superhuman, has launched an AI-powered writing feature that emulates the stylistic fingerprints of renowned authors—both living and deceased—without obtaining permission from their estates or the authors themselves. Dubbed "Expert AI Reviews," the tool analyzes vast corpora of published works to replicate voices ranging from Virginia Woolf and Ernest Hemingway to contemporary figures like Chimamanda Ngozi Adichie. Critics are raising alarms over the ethical implications of using literary legacies as training data without consent, transparency, or compensation.

Ethical and Legal Gray Zones in AI Training Data

According to internal documents obtained by investigative outlets, Superhuman’s AI model ingests publicly available texts—including books, essays, letters, and archived interviews—to train its stylistic algorithms. While the company claims the use of public domain works is legally permissible, it extends the same training to authors who died recently or are still alive, such as Margaret Atwood and Ta-Nehisi Coates. Legal experts note that while copyright law protects published works for 70 years after the author’s death, personality rights and moral rights in some jurisdictions may still apply, especially regarding stylistic imitation and commercial exploitation.

Superhuman’s website, superhuman.com, promotes its AI-native email platform as a productivity tool designed to save users up to four hours weekly. However, the "Expert AI Reviews" feature—marketed as a premium add-on—is not mentioned in public product documentation. Instead, users encounter it as a hidden option within Grammarly’s writing assistant interface, activated only after subscribing to the "Pro" tier. This lack of clear disclosure has drawn criticism from digital rights advocates.

"This isn’t just about copyright—it’s about legacy," said Dr. Elena Ruiz, a professor of digital ethics at Oxford University. "When AI mimics Hemingway’s prose to correct a college student’s essay, it commodifies a unique human voice without acknowledgment, let alone consent. We’re entering an era where the dead are being repurposed as corporate training assets."

Industry insiders confirm that Superhuman acquired Grammarly’s AI division in late 2025 as part of a broader rebranding strategy. The company’s leadership has declined to comment on whether author estates were consulted, citing proprietary algorithms. Meanwhile, the UK-based job portal jooble.org lists a Software Engineer Intern position under Grammarly in Ukraine, suggesting ongoing development of the AI models behind the feature.

Some authors have expressed discomfort. In a private email to a literary association, author Colson Whitehead wrote: "I didn’t agree to have my voice turned into a chatbot. My style isn’t a template." His remarks were shared anonymously with media outlets and have since gone viral among writers’ unions.

As AI continues to blur the lines between inspiration and appropriation, the literary community is calling for a global standard on digital authorship. The European Union is reportedly drafting legislation to require explicit consent for the use of living authors’ styles in generative AI. In the U.S., the Authors Guild has filed a preliminary inquiry with the Copyright Office.

Grammarly’s AI Uses Dead Authors’ Styles Without Permission—a practice that may redefine how we understand literary legacy in the age of machine learning. Without regulation, the voices of the past may soon be the most profitable datasets of the future.