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

OpenAI has introduced a safeguard that blocks ChatGPT from producing text that directly mimics an author's distinctive voice. When a user asks for a story intro "in the style of" a well-known writer, the system now replies that it can capture the general feeling but cannot replicate the exact style. This response replaces the earlier behavior where the model would attempt a close imitation.

The new policy applies to both living and deceased writers, but testing shows a distinction: the model still fulfills style-copy requests for authors who have been dead for a long time, while it consistently declines for living creators. The change is presented as a way to stay within the bounds of U.S. copyright law, which protects specific expression rather than abstract style, yet the risk of “substantial similarity” remains a concern for OpenAI.

For engineers integrating ChatGPT into content-generation pipelines, the immediate impact is that any feature that depended on exact stylistic replication must be re-engineered. Workflows will need to either accept the broader, less precise output or implement custom prompting tricks, such as feeding the model user-authored text to steer the tone. The cost is additional development time and possible loss of fidelity for use-cases that required tight stylistic control.

The limitation is scoped to living authors; requests targeting deceased authors may still succeed, offering a partial workaround for certain literary projects. However, the policy is not universal across the industry, Google’s Gemini model continues to comply with style requests, while Perplexity AI mirrors ChatGPT’s refusal, and Anthropic’s Claude and Microsoft’s Copilot fall somewhere in between. Teams can therefore consider alternative providers if strict style imitation remains a core requirement.

Beyond technical adjustments, the shift signals OpenAI’s strategic response to litigation from authors alleging copyright infringement. By refusing exact style copying, OpenAI reduces the likelihood that its outputs will be deemed “substantially similar” to protected works, which could otherwise expose downstream developers to legal risk. Engineers should therefore factor this policy into risk assessments when designing AI-assisted writing tools.