OpenAI’s ChatGPT has updated its guidelines to block direct requests for text that mirrors famous authors’ styles. Instead, the model now generates content with a nod to those ‘broad qualities’ while remaining distinctly its own.

This morning, when asked to emulate Stephen King, ChatGPT provided an original opening that captured a similar feeling while maintaining its unique voice. The response reflected a broader approach, offering an 'atmospheric, character-driven horror' without directly copying King's exact style.

The move comes as OpenAI faces legal battles over copyright infringement claims by authors whose works have been used to train AI models. By refusing direct style copying, ChatGPT aims to skirt the line between inspiration and imitation, potentially avoiding accusations of infringing on original expression.

In the US, copyright law protects specific expressions but not the overall style of an author. However, if the AI-generated text becomes 'substantially similar' to a copyrighted work, it could still fall foul of infringement laws.

“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,” explained George Washington University Law School Professor Robert Brauneis. As AI continues to evolve, the fine line between homage and theft remains under scrutiny.