ChatGPT now refuses to copy a living author's writing style
OpenAI's chatbot has quietly started blocking requests to mimic named authors, offering a watered-down alternative instead. Here is what changed and why it matters.
Key points
- ChatGPT now refuses direct requests to copy the writing style of named authors, offering a blended alternative instead.
- In testing, the block applies to both living authors (J.K. Rowling, Amy Tan) and deceased ones (Charles Dickens, Ernest Hemingway).
- An earlier analysis by No Latency found the restriction applied only to living authors, suggesting the policy may still be evolving.
- OpenAI has not issued a public statement explaining the change.
Ask ChatGPT to write something "in the style of Stephen King" and you will no longer get what you asked for. The chatbot, a large language model (software trained on vast amounts of text that can write, answer questions and hold conversations), now sidesteps those requests entirely.
Instead of mimicking a specific author, it produces something that draws on what it calls the "broad qualities" of that writer while keeping its "own voice." Ars Technica tested the behaviour this week and received this reply to a Stephen King prompt: "I can definitely write with the hallmarks of atmospheric, character-driven horror and small-town dread, but I can't write in Stephen King's exact style or closely imitate his distinctive voice."
Why would OpenAI do this?
Style itself is not protected by copyright law, but closely imitating a recognisable author raises real legal and ethical questions that AI companies are still working through. Reproducing someone's distinctive voice too faithfully could blur the line between inspiration and imitation in ways that courts have not yet settled.
OpenAI has not explained the change publicly. No statement, no blog post.
The block covers a broad range of names. Ars Technica confirmed it triggers for living writers including J.K. Rowling and Amy Tan, as well as long-dead ones like Charles Dickens and Ernest Hemingway. That last detail is notable: copyright on those authors' work has long expired, so the refusal appears to go beyond strict legal caution.
Does this match what researchers have already found?
Partly, but not entirely. Analysis published by No Latency earlier this month found the same restriction for living authors. For deceased authors, however, No Latency found ChatGPT would still comply. The gap between those findings and current behaviour suggests OpenAI may have widened the restriction recently, possibly more than once.
The picture is still shifting.
What does this mean for writers and creatives?
For most people using ChatGPT to draft stories or experiment with different tones, the practical impact is modest. The chatbot still writes fiction, still adjusts its register, and still captures genre conventions. What it will not do is produce output that reads like a specific named person wrote it.
For professional writers, that is arguably the point. Authors have long complained that AI tools trained on their work now compete with them commercially. A policy that stops the tool from wearing a named author's face directly is a small but visible concession to those concerns.
Watch for: prompts that name an author being silently redirected rather than refused outright. ChatGPT does not always say no; sometimes it just quietly produces something different without flagging the change. If your output feels generic when you expected a specific voice, the new policy may be the reason.
