ChatGPT is backing away from one of generative AI’s flashiest tricks: sounding exactly like somebody else. The shift looks small on the surface, but it lands right where OpenAI is most exposed — in the widening fight over copyright and creative ownership.

What’s changed is the wording, and possibly the strategy. Reports say ChatGPT now declines blunt requests to write in the style of famous authors, instead offering something that borrows the “broad qualities” or general feel of a writer without copying a “distinctive voice.” Ars Technica found the bot now blocks direct style-cloning requests and pivots to language about staying “distinct in its own voice.”¹ The Verge reported the same broader pattern, noting that ChatGPT “now refuses to mimic famous authors.”²

That may sound like a semantic dodge. But legally, semantics are the whole game. U.S. copyright law traditionally protects expression, not an author’s abstract style. Even so, the line gets blurry when a machine can produce something close enough to feel copied. That is why this new refusal matters: OpenAI appears to be drawing a cleaner boundary between inspiration and imitation just as author lawsuits keep pressing the claim that AI systems can reproduce copyrighted work too convincingly.¹

There’s also a practical split in how people will read this change. To critics, it looks like overdue restraint after years of training on creative labor without clear permission. To defenders of AI tools, it is a sensible guardrail — one that still allows users to ask for atmosphere, pacing, or genre traits without demanding a synthetic Stephen King or Agatha Christie.

Either way, the message is unmistakable. OpenAI is not merely tweaking product behavior; it is narrowing a legal risk surface. “ChatGPT starts blocking direct requests to copy an author's style” is more than a product note.¹ It is a sign that the industry’s anything-goes mimicry era may be getting harder to defend.