Requests for ChatGPT to copy the style of living authors now trigger refusals. The AI offers original writing that echoes certain qualities but avoids direct imitation. Legal risks and user frustration are rising.
Media and publishing professionals relying on AI-generated content are facing new restrictions as OpenAI’s ChatGPT now declines to directly imitate the writing style of well-known living authors. Instead, the platform produces original text that reflects some general characteristics of the requested author’s work, but stops short of close stylistic mimicry. This shift is already impacting editorial workflows, content production strategies, and the ability to generate traffic-driving material tailored to specific voices.
Recent tests show that when prompted to write in the style of authors such as Stephen King, J.K. Rowling, or Amy Tan, ChatGPT responds by referencing the broad qualities of their writing-such as tone or atmosphere-while explicitly refusing to replicate their distinctive voice. The model’s responses for deceased authors like Charles Dickens or Ernest Hemingway have been inconsistent, with some studies reporting that ChatGPT still complies with style-copying requests for those writers. This nuanced approach appears to be a response to ongoing legal challenges, as OpenAI faces lawsuits from book authors who allege copyright infringement by AI models trained on their work.
Legal experts note that while US copyright law protects specific expressions rather than an author’s style, AI-generated imitations could cross into infringement if they become substantially similar to original works. The Authors Guild has advised writers to avoid using generative AI to copy or mimic the unique styles of fellow authors, warning of both ethical and legal risks. High-profile incidents, such as the controversy surrounding Lena McDonald’s book that included AI-generated passages mimicking another author, have fueled debate about the boundaries of AI-assisted creativity and fair competition.
Not all large language models handle style imitation requests the same way. According to a study by No Latency, Google’s Gemini model continues to comply with requests to mimic author styles, while Perplexity AI’s models consistently refuse and redirect users, similar to ChatGPT. Anthropic’s Claude and Microsoft’s Copilot reportedly offer qualified compliance, acknowledging the imitation issue. OpenAI’s DALL·E 3 image generator already blocks requests to create images in the style of living artists, but the company’s written content guidelines remain less explicit. These inconsistencies are creating confusion for users who depend on AI tools for content creation, with some expressing frustration over the new limitations and seeking workarounds to maintain their editorial output.
For publishers and creators, these changes echo broader shifts in how AI tools are regulated and deployed across the industry. As seen in other sectors, such as sports media’s use of digital products to drive engagement-like the approach detailed in this report on local sports coverage innovation-the evolving capabilities and restrictions of AI platforms are forcing content teams to adapt their strategies for audience growth and monetization.
OpenAI, founded in 2015, has rapidly become a central player in the generative AI market. The company’s ChatGPT platform reached over 100 million weekly active users by 2024, according to official figures. OpenAI’s ongoing legal battles and policy adjustments reflect the high stakes involved as AI-generated content becomes increasingly integrated into publishing, media, and creative industries worldwide.
