Skip to Content

OpenAI will add an invisible watermark to ChatGPT text in the EU

On October 5, OpenAI explained how it will comply with the AI Act requirement that generated text be identifiable in a machine-readable way.
October 5, 2026
Curated High

What it says

  • In the coming weeks it will add an invisible watermark to ChatGPT and Codex text for eligible users on all plans, only in the European Union.
  • In the API, customers worldwide can already enable it on some models; by default it remains disabled.
  • The technology, textGrain, introduces a statistical signal into word choice. At first, the detector will be available only to approved researchers and expert organizations.
  • Its limits, according to OpenAI itself: with 1% false positives, it detects around 80% of 200-token passages and 95% of 400-token ones. Replacing 10% of the words lowers detection from 92% to 66%; replacing 25%, to 17%.
  • OpenAI stresses what the mark does not say: it does not measure human contribution, does not identify the user or establish authorship, and its absence does not prove that a text is human.

How it has evolved

Thread: Application of the AI Act. How the European AI Act is applied in practice: timeline, amendments, Commission guidelines, Spanish adaptation, and providers' response. Upcoming dates: December 2, 2026 (Article 50.2 marking for systems already on the market) and December 2, 2027 (Annex III high risk).

The Tech Guardian summarizes and comments in its own words on third-party publications. The full information belongs to the original source, which we recommend you read.

Our take

If proposals, reports or client messages are drafted with ChatGPT in your company, those texts will carry a detectable signal, even if today only authorized researchers can check it. That is no reason to stop using it, but it is a reason to be clear about your policy: what is reviewed, who signs it and what the client is told. The transparency you will be asked for will concern that process, not the brand.

Spain's data protection agency warns a company that wanted to score résumés with AI: these are the safeguards
The Spanish Data Protection Agency (AEPD) made public on September 23 a warning to a company that was preparing an AI tool to screen job applications.