ChatGPT now refuses to imitate famous authors' writing styles, offering mood and themes instead

OpenAI's ChatGPT has quietly changed its behavior: it now refuses direct requests to imitate the writing style of famous authors, according to reports from Ars Technica and independent testing. Instead of producing a close imitation, the chatbot offers to capture the broad mood, themes, or storytelling qualities of the requested author. This shift has immediate implications for AI builders who rely on style-mimicry features in writing assistants, content generators, or creative tools.

What happened

ChatGPT now refuses to copy the voices and distinctive writing styles of popular authors, as documented by Ars Technica and echoed by No Latency. In a test involving Agatha Christie, ChatGPT replied that Christie's works remain under copyright and it cannot imitate her distinctive style, while still offering to write something with those characteristics. The same refusal applies to both living and dead authors, including Stephen King, J.K. Rowling, and Charles Dickens. When asked for a horror story in King's style, ChatGPT responded that it can write with the hallmarks of atmospheric horror but not imitate his exact voice. The change appears to be a policy-driven tightening to reduce copyright risk, as OpenAI faces multiple lawsuits over training data.

Why AI builders should care

If your product uses ChatGPT to generate content in a specific author's voice, this change directly affects your prompt design and user experience. You can no longer rely on simple prompts like "write in the style of [author]" to produce acceptable output. This matters for any team building writing assistants, content generators, or creative tools that depend on style-mimicry. The policy signals that other AI providers may adopt similar restrictions, especially as copyright litigation continues. Builders should plan for a future where style imitation is explicitly governed, and design prompts that avoid direct voice copying while still achieving the desired tone or genre.

Practical implications

Prompt design: Avoid asking for exact imitation of an author's voice. Instead, describe the mood, genre, pacing, or structural elements you want. For example, ask for "a mystery with a clever detective, red herrings, and a twist ending" instead of "write like Agatha Christie." This approach is more likely to comply with the new policy while still producing useful output.

Compliance and attribution: If your use case requires capturing a specific author's voice, consider using a model that doesn't have these restrictions, or fine-tune on uncopyrighted or public domain content. But be aware of the legal risks: direct imitation of copyrighted works may invite lawsuits.

Soft block: ChatGPT's refusal is not a hard ban in all cases. It may still produce text with similar qualities, suggesting the change is a discretionary disclaimer. However, this could tighten over time, so treat it as a signal to update your prompt templates now.

Caveats

This is a recent change, and details may evolve as OpenAI updates its policies. The refusal is not absolute: ChatGPT still offers to write with the broad qualities of an author's style, so the boundary is fuzzy. The behavior has been observed on the consumer ChatGPT interface; API behavior may differ. Builders should test their own prompts and monitor for further changes. The evidence is based on reports from Ars Technica, No Latency, and Engadget, and may not reflect every model or version.