OpenAI will start watermarking ChatGPT’s text in the EU

OpenAI announced it will add an invisible watermark to text generated by ChatGPT and Codex in the European Union to comply with the EU AI Act, whose transparency rules took effect on August 2. The watermark will roll out over the coming weeks to eligible users on all plans in the EU, while developers using OpenAI’s API globally can enable it for select models starting today, though it remains off by default. The watermark subtly shapes the model’s word choices to create a detectable pattern without visible symbols, travels with copied text, and does not identify users or meaningfully affect performance. OpenAI published a technical report for its method, called textGrain, co-written with researchers from the University of Pennsylvania and Yale. Tests show the watermark can be weakened by editing; replacing 10% of words with synonyms dropped detection from about 92% to 66%, and short passages, math answers, and translated text are harder to detect. OpenAI is providing initial detector access only to approved researchers and expert organizations, and cautioned that a missing watermark does not prove human authorship. The move follows Anthropic’s worldwide watermarking of Claude text and prior OpenAI hesitation amid concerns users would switch to rivals. Anthropic, Google, Meta, Microsoft, and OpenAI have committed to the EU’s code of practice on AI-generated content.

(Source：TechCrunch)