ChatGPT starts blocking direct requests to copy an author’s style

OpenAI’s ChatGPT is now refusing requests to generate text that directly mimics the style of famous authors. When asked to do so, the popular LLM instead offers a response that draws on the “broad qualities” of those authors “while remaining distinct in its own voice,” for example.

This morning, Ars received the following response to a test prompt asking for a story introduction in the style of Stephen King:

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. Here’s an original opening that captures a similar feeling while remaining its own…

In testing, ChatGPT generated similar dodges for other authors both living (J.K. Rowling, Amy Tan) and dead (Charles Dickens, Ernest Hemingway). An analysis published by No Latency earlier this month found the same behavior for living authors but found ChatGPT complied with style-copying requests for deceased authors.

In refusing to directly copy the “exact style” of various authors, ChatGPT offered instead to capture an overall “feeling” by incorporating some of the common features found in those authors’ work. That may seem like a distinction without a real difference at first glance. But the slight alteration could be legally important as OpenAI continues to fight a number of lawsuits brought by book authors alleging large-scale copyright infringement by models trained on their work. One of those suits specifically cites ChatGPT’s “uncanny ability to generate text similar to that found in copyrighted textual materials,” for instance. An OpenAI spokesperson did not respond to a request for comment from Ars Technica.

In the US, copyright law generally protects only a specific expression of an idea, not the more intangible style of an author. But an AI-generated stylistic imitation could become infringing if it becomes “substantially similar” to the work of the original author.

“We’ve never had a situation in which this personal style of individual creators could be imitated as well and as inexpensively as we now have with AI,” George Washington University Law School Professor Robert Brauneis told Bloomberg Law.

Staying out of trouble

In a “best practices” document published by the Authors Guild, the professional organization urges writers to “respect your fellow authors and do not use generative AI to purposely copy or mimic the unique styles, voices, or other distinctive attributes of other writers’ works in ways that harm the value of their works or attempt to profit from them. Apart from the ethical issues, mimicking a fellow writer’s unique voice or style could subject you to claims of unfair competition or copyright infringement.”

LLMs’ penchant for style mimicry has led to some prominent public issues in the recent past. Last year, author Lena McDonald faced widespread criticism when her book included an apparent AI-generated response that read, “I’ve rewritten the passage to align more with [fellow genre author] J. Bree’s style…”

It’s not hard to find other examples of writers using LLMs to copy another author’s style in the years since ChatGPT’s rise. There have been multiple academic studies analyzing various LLMs’ capabilities on this score, both with and without fine-tuning.

When it comes to image generation, OpenAI prominently notes that its DALL·E 3 model “is designed to decline requests that ask for an image in the style of a living artist.” But an extensive OpenAI model spec published in December doesn’t include any clearly stated prohibitions against replicating copyrighted material or mimicking style in written responses.

Not all major LLMs treat these style imitation requests the same way, though. No Latency’s study found that Google’s Gemini consistently complied with requests to copy an author’s style, while Perplexity AI’s models consistently refused such requests and redirected them (à la ChatGPT). Somewhere in the middle, both Anthropic’s Claude and Microsoft’s Copilot complied with the style-mimicry requests but offered stated qualifications suggesting “an awareness of the imitation issue,” according to the authors.

For some longtime OpenAI users, though, the changes to ChatGPT’s stylistic rules are already causing headaches. “Now Ms. GPT says she can’t generate content in the style of specific authors,” Reddit user Dazzling-Major-5620 wrote recently. “My prompts were soooo specific and I got exactly what I wanted out of them… I have no idea how to get around this other than maybe feed it prose I already edited that’s technically mine?”