OpenAI announced on Monday that it will begin adding an invisible watermark to text generated by ChatGPT and Codex in the European Union. This move is in compliance with the EU AI Act, which mandates transparency rules that took effect on August 2. These rules require AI companies to mark AI-generated content in a way that can be identified by other systems.

Watermark Rollout

The watermark will be rolled out over the coming weeks to eligible ChatGPT and Codex users on all plans, but only within the EU. Developers using OpenAI’s API globally can enable it for select models starting today, although it is off by default. OpenAI clarified that it is not making text watermarking a global default at launch.

How the Watermark Works

The watermark is not a visible symbol; instead, it subtly shapes the model’s word choices, creating a pattern that is undetectable to readers but can be identified by a detector. Since it is embedded in the words themselves, it remains intact when the text is copied and pasted. OpenAI emphasized that the watermark does not identify the user and that there was no significant change in the performance of its models with the watermark activated.

OpenAI also released a technical report detailing its method, called textGrain, co-authored with researchers from the University of Pennsylvania and Yale. This report explains how a secret key can be used to influence next-word predictions to complete sentences. By combining numerous subtle adjustments, the detector can identify AI-generated content using just the text and the key.