OpenAI Unveils New AI Text Watermarking System Coming to ChatGPT, Codex, and the API

OpenAI is rolling out a new way to mark its generated text as a direct response to a European regulation that requires AI companies to make their output identifiable.The rollout touches ChatGPT, Codex, and the company's developer API, though each one is being handled with a different level of urgency and availability.OpenAI Unveils New Watermarking SystemOpenAI announced a new text watermarking technology called textGrain, built in response to the EU AI Act's requirement that generative AI providers make machine-generated text identifiable in a machine-readable way. The company described the rollout as a phased approach, acknowledging upfront that text watermarking and detection remain early technologies with real limitations.TextGrain works by adding an invisible statistical signal to the model's word choices as it generates a response.