OpenAI has updated ChatGPT to block direct requests that ask the AI to copy the writing style of well-known authors. When prompted to generate text in the voice of figures like Stephen King, J.K. Rowling, Amy Tan, Charles Dickens, or Ernest Hemingway, the model declines to replicate their "exact style" and instead offers original content inspired by broad stylistic traits such as atmospheric horror or character-driven storytelling. A test by Ars Technica confirmed the change, with ChatGPT providing a disclaimer and an alternative passage that evokes a similar mood without direct imitation.
This shift follows legal pressure from authors who have filed lawsuits alleging that AI models like ChatGPT were trained on copyrighted books without permission. One lawsuit specifically points to the system's ability to produce text "similar to that found in copyrighted textual materials" as evidence of infringement. While U.S. copyright law does not protect an author's style alone, it does guard against expressions that are "substantially similar" to protected works. By distancing itself from direct replication, OpenAI appears to be navigating this legal boundary. Earlier analysis by No Latency showed that ChatGPT previously complied with style mimicry requests for deceased authors, but current behavior indicates a broader policy change across both living and dead writers.
The refusal to copy author styles suggests OpenAI is adjusting its output to reduce legal exposure, not to uphold creative ethics. This matters because the line between "inspired by" and "substantially similar" remains untested in court, leaving authors and AI firms in a gray zone. If the model still produces text that echoes protected expression, the policy change may be more about optics than legal safety.
Editorial note: AI-assisted opinion, not established fact. Full disclaimer →
