OpenAI to add watermarks to ChatGPT in EU
Published: 05/10/2026
| TechCrunch
In a blog article, OpenAI announced it will introduce invisible text watermarks to content generated by ChatGPT and Codex within the EU to comply with transparency obligations under the Artificial Intelligence Act (AI Act). The rules, which came into effect on 2 August, require developers to mark artificial intelligence (AI) outputs so automated systems can detect them.
The watermarking system, detailed in a technical report developed with researchers from the University of Pennsylvania and Yale, subtly adjusts next-word selections using a secret cryptographic key. These adjustments create patterns invisible to human readers that move with copied text without identifying the individual user or reducing model performance.
The feature will roll out over the coming weeks for EU users across all subscription tiers, while developers using the company's API in other parts of the world can manually activate it for selected models.
OpenAI noted that detection rates fall significantly when passages are short, translated, or edited with synonyms, prompting the company to restrict initial detector access to approved researchers and organisations. It also noted that the absence of a watermark does not verify human authorship, nor does its presence reflect the level of human creativity involved.
The rollout follows a global watermarking deployment by rival Anthropic in August.
Training Announcement: Freevacy offers a range of independently recognised professional AI governance qualifications and AI Literacy short courses that enable specialist teams to implement robust oversight, benchmark AI governance maturity, and establish a responsible-by-design approach across the entire AI lifecycle. Find out more.
Image from Shutterstock, credit StepanPopov
What is this page?
You are reading a summary article on the Privacy Newsfeed, a free resource for DPOs and other professionals with privacy or data protection responsibilities helping them stay informed of industry news all in one place. The information here is a brief snippet relating to a single piece of original content or several articles about a common topic or thread. The main contributor is listed in the top left-hand corner, just beneath the article title.
The Privacy Newsfeed monitors over 300 global publications, of which more than 3,250 summary articles have been posted to the online archive dating back to the beginning of 2020. A weekly roundup is available by email every Friday.