OpenAI announces text watermarking for ChatGPT in the EU
OpenAI has announced textGrain, an invisible statistical watermark for AI-generated text, with a rollout to eligible ChatGPT and Codex users in the European Union over the coming weeks. The announcement was published on 5 October.
The EU rollout covers eligible users across plans. Separately, API developers worldwide can opt in on selected models. Access to the detector will initially be limited to approved researchers and expert organisations, rather than being publicly available.
OpenAI says the watermark has limits. An absent watermark does not prove that text was written by a person, and a detected watermark cannot measure the extent of human contribution. Editing, translation, short passages and unsupported models can reduce reliability.
For schools, this is a development to watch rather than a ready-made assessment policy. Conversations about students’ work, evidence of the writing process and clear expectations for AI use remain useful. Any future detector would need to be understood on its own terms before its results were used to make decisions about a student.
Comments