OpenAI launches textGrain text watermarking with staged rollout
What happened
On October 5, OpenAI published textGrain, a text watermarking scheme meant to address the EU AI Act's text provenance rules, and began a staged rollout. Starting that day, API customers worldwide can turn watermarking on for some models; the feature is off by default. Under the published plan, eligible ChatGPT and Codex text output in the EU will carry an invisible watermark in the coming weeks. The matching detector is initially open only to approved researchers and professional organizations by application. For now the API opt-in is available, while deployment in EU products remains a later step.
Written by AI from the coverage · updated 1 hour ago
Coverage
Follow the reports to see the story from different sides.
- OpenAI NewsPickOpenAI's textGrain watermark: phased rollout to meet EU text provenance rules
OpenAI published textGrain, its text watermarking plan for the EU AI Act. From today, API customers worldwide can turn it on for some models; it stays off by default. In the coming weeks, eligible ChatGPT and Codex text output in the EU will carry an invisible watermark, and the detector opens to approved researchers and professional organizations by application only.
Heat over time
Not enough continuous observations to draw a trend yet.