Overview
- Independent testing reported last week showed ChatGPT now declines direct requests to write in a named author’s voice and instead offers an original piece that draws on broad qualities of that writer.
- When prompted for a specific author the model gives a polite refusal followed by an alternative that evokes general themes or tones rather than copying precise sentence‑level style.
- The change follows multiple lawsuits filed by authors since 2023 who say their books were used without permission to train large language models and comes after a large copyright settlement by a rival AI firm.
- Developers and crypto or Web3 projects that built features around precise ‘author-style’ generation must reassess those features because model behavior has shifted even though no crypto tokens are directly tied to the update.
- This move echoes OpenAI’s earlier pause on voice‑cloning features and signals that court rulings or licensing deals will likely shape whether the restriction becomes permanent across the industry.