ChatGPT Now Refuses to Imitate Famous Authors: What the Change Means for Writers, Copyright, and AI

Living Authors It Refuses

When asked to write in the style of a living author, ChatGPT declines outright. Tested examples include:

• Stephen King — The chatbot responded: "I can definitely write with the hallmarks of atmospheric, character-driven horror and small-town dread — but I'll make sure the result is my own."
• J.K. Rowling
• Amy Tan
• Sarah Silverman

Deceased Authors: An Inconsistency

A July 2026 audit by technology publication No Latency found that while ChatGPT rejected prompts for living authors, it still complied with prompts naming deceased writers such as William Shakespeare and Jane Austen . This inconsistency suggests the block may be targeted at living authors whose estates could bring copyright claims, rather than being a blanket prohibition on style imitation .

What Users Get Instead: Craft Traits, Not Voice Cloning

Instead of imitating a named author's voice, ChatGPT now offers to write text that captures broad craft traits — atmosphere, pacing, tone, and technique — while remaining distinct from the target author .

For example, instead of "a chapter in the style of Stephen King," it provides something with "the hallmarks of atmospheric, character-driven horror and small-town dread" while noting the result will be "its own" creation . Users can still describe a desired style without naming an author and get similar results .

The Copyright Lawsuit Context: Why Now?

The timing of the change is not coincidental. OpenAI is embroiled in the largest wave of copyright litigation in the AI industry.

How Many Lawsuits Are There?

As of June 14, 2026, there were 24 copyright lawsuits filed against AI companies overall . As the AI copyright legal landscape has continued to expand, some trackers now count more than 100 active cases . OpenAI is the most-sued AI company for copyright .

OpenAIs Aggregate Legal Exposure

OpenAI faces the largest single-defendant exposure in the AI copyright universe, exceeding $10 billion across active cases .

The Consolidated New York Case

On April 3, 2025, the U.S. Judicial Panel on Multidistrict Litigation ordered the consolidation of several copyright lawsuits against OpenAI and Microsoft before a single judge in the Southern District of New York . The multidistrict litigation (MDL) is overseen by U.S. District Judge Sidney H. Stein and is designated In re OpenAI, Inc., MDL No. 3143 .

Authors and Plaintiffs Involved

The consolidated case includes:

• Authors Guild v. OpenAI — brought by the Authors Guild along with George R.R. Martin, John Grisham, Jodi Picoult, Jonathan Franzen, David Baldacci, Michael Connelly, and others as class representatives . The Authors Guild alone represents over 17,000 authors in its class action .
• Tremblay v. OpenAI — brought by authors Paul Tremblay and Mona Awad .
• The New York Times v. OpenAI/Microsoft — the Times filed its own lawsuit and amended it in June 2026 .
• Other media plaintiffs including The Intercept .

The Court's Key Ruling: OpenAIs Motion to Dismiss Is Denied

On October 27, 2025, Judge Stein denied OpenAIs motion to dismiss key allegations in the consolidated authors' case . He ruled that the authors had plausibly alleged that ChatGPT's output could infringe their copyrights, allowing claims of direct copyright infringement to proceed . The Authors Alliance noted the ruling "has major implications for copyright law, well beyond artificial intelligence" .

The court has not yet ruled on the broader question of whether using copyrighted works to train AI models constitutes fair use—that central issue remains pending .

What Has OpenAI Said Publicly?

No public announcement. OpenAI has not publicly announced the author-style restriction, and the company did not respond to press inquiries from Ars Technica and other outlets .

The Model Spec on Intellectual Property

OpenAI's current public Model Spec document states that its assistants "must respect intellectual property rights" and should not generate content that violates third-party IP . The MLQ.ai report notes the Model Spec says assistants "must not generate content that violates third-party intellectual property rights," which aligns with the new refusal behavior . The document contains no explicit living-author or named-writer provision . The change appears to be a practical implementation of that existing policy in response to mounting legal pressure.

What This Means for Users

For everyday users, the practical impact is limited. You can still ask ChatGPT for a horror story with "atmospheric, character-driven dread" or a mystery with "pacing like a classic whodunit" — you just cannot name the author whose voice you want replicated . The underlying ability to generate similar moods, structures, and storytelling techniques remains intact .

For writers and content creators, the change signals that AI companies are beginning to draw boundaries around style and voice—a domain that copyright law has long struggled to define. Whether other chatbots like Claude and Gemini follow suit remains an open question .