Policy · Updated 6 Oct, 02:09 pm IST
OpenAI adds text watermarking to ChatGPT and Codex in the EU
Why it matters for readers: It's a way companies are trying to make AI output traceable, though the marks aren't perfect and don't certify who wrote the text.
- OpenAI will introduce an invisible statistical watermark called textGrain for ChatGPT and Codex outputs for eligible users in the EU first.12
- The company says textGrain matched or exceeded other approaches like Google DeepMind’s SynthID and Anthropic’s watermarking in its internal comparisons.1
- API customers worldwide can opt in to receive watermarked text outputs for select models starting immediately.1
- OpenAI cautions that detection can weaken when text is short, constrained, or heavily edited, and that watermarks do not verify accuracy, authorship, or human contribution.13
- Watermarking will not be made a global default at launch; OpenAI will limit the initial rollout to the EU to gather feedback.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get startedHow this story developed
5 Oct, 11:38 pm · The Verge AI
OpenAI is adding text watermarking in ChatGPT and Codex
6 Oct, 02:06 am · TechCrunch AI
OpenAI will start watermarking ChatGPT’s text in the EU
6 Oct, 11:59 am · MediaNama
OpenAI to watermark AI-generated text in EU, but detection has limits
6 Oct, 02:09 pm · Indian Express Technology
OpenAI is adding watermarks to ChatGPT text in the EU: Here’s what it means