OpenAI tells ChatGPT to stop impersonating famous authors

Details

OpenAI has issued an internal directive instructing ChatGPT to refuse requests that ask it to impersonate specific famous or living authors. The move is a direct response to sustained pressure from author organizations, publishers, and individual writers who argue that AI systems capable of closely mimicking a writer's distinctive voice threaten their livelihoods and intellectual property rights. The policy sits at the intersection of copyright law and AI content policy. While generating text 'in the style of' an author has generally been treated as legally ambiguous, producing output indistinguishable from a named author's actual work raises clearer concerns around commercial harm and identity. OpenAI appears to be drawing a harder line than it previously held. For practitioners building content generation pipelines on ChatGPT, the behavioral change may break prompts that relied on author-voice instructions. It is a guardrail update, not a model capability change — broad stylistic prompting likely still works, but explicit impersonation requests will be declined. The directive fits a broader pattern of AI labs implementing self-regulatory measures on copyright-sensitive features ahead of expected legislation. The source article is thin on operational specifics — exact enforcement mechanisms and which authors are covered were not detailed.