OpenAI to watermark ChatGPT and Codex text in the EU under the AI Act

OpenAI said Oct. 5 that it will add an invisible watermark, called textGrain, to eligible ChatGPT and Codex text output in the European Union over the coming weeks, in response to the EU AI Act’s requirement that generative AI providers make generated text identifiable by machines; API customers worldwide can opt in for select models, with watermarking off by default, and access to its detector is limited at first to approved researchers and expert organizations. By OpenAI’s own tests, at a 1% target false-positive rate the detector found watermarks in about 80% of 200-token passages and about 95% of 400-token passages, and replacing 10% of the words in 400-token passages with synonyms cut detection from about 92% to 66%, while replacing 25% cut it to 17%. OpenAI says a detected watermark does not identify the user or establish who is responsible for the text, and that the absence of one does not prove human authorship; The Register noted that Anthropic, unlike OpenAI, applies its text watermarking outside Europe too.