ChatGPT Blocks Requests to Imitate Famous Authors' Writing Styles
OpenAI tightens copyright guardrails as AI faces mounting legal pressure from publishers and authors.
ChatGPT now refuses direct requests to mimic the writing styles of famous authors, whether living or deceased, marking a notable shift in how OpenAI handles intellectual property concerns.
A Sudden Policy Shift
Testing by Ars Technica confirmed the change across multiple authors including Stephen King, J.K. Rowling, Amy Tan, Charles Dickens and Ernest Hemingway. When prompted to copy a specific author's style, the AI now cites copyright concerns and declines.
According to research from No Latency, referenced in Ars Technica and RuntimeWire reporting, ChatGPT previously complied with imitation requests for deceased authors. The expanded refusals appeared between July 15-27, 2026, suggesting a deliberate policy change rather than gradual model drift.
Copyright Cited as Primary Concern
In documented exchanges, ChatGPT explicitly references copyright status when declining requests. Ars Technica recorded the system refusing Stephen King imitation on copyright grounds.
The AI does offer a compromise: it will write content capturing the general "feeling" or common characteristics of an author's work while remaining original. "I can certainly write something with those characteristics, while remaining original," the system told users in tested interactions.
Legal Pressure Mounts
The timing aligns with escalating legal challenges facing OpenAI. The company currently defends against multiple copyright lawsuits from Encyclopedia Britannica, Merriam-Webster, and the Authors Guild, among others.
These cases center on whether training AI models on copyrighted works constitutes fair use, and whether output that replicates distinctive creative voices infringes on intellectual property rights.
Drawing a Line
This behavioral change represents OpenAI attempting to draw a defensible boundary between learning from training data and replicating a specific creator's identifiable voice. By refusing to "clone" named authors while offering generic stylistic approximations, the company signals awareness that exact imitation carries legal risk that broader stylistic influence may not.
For writers and publishers, the shift validates longstanding concerns about AI systems commodifying distinctive creative voices. For OpenAI, it represents a pragmatic retreat from a legally precarious position as copyright litigation proceeds through the courts.
The change affects all users equally—no workaround exists for imitating protected authors, regardless of their living status. The guardrail appears hardcoded rather than prompt-sensitive, suggesting enforcement at the model or system instruction level.