OpenAI will start watermarking ChatGPT text in the EU

OpenAI announced it will begin watermarking text generated by ChatGPT and Codex for users in the European Union, a move tied to compliance with the EU’s AI Act. According to TechCrunch, the watermarks will be invisible to readers but detectable through technical means.

The company stated that the marks are not foolproof. OpenAI said editing or substantially altering the generated text can make the watermark harder to detect, reducing its reliability as a universal signal of AI authorship.

TechCrunch frames the rollout as part of a broader shift occurring under the AI Act, which is pushing companies from voluntary disclosure practices toward mandated ones. Where AI firms previously decided independently whether and how to flag machine-generated content, the regulation now requires specific transparency measures for systems operating in the EU.

The report does not specify an exact rollout date for the watermarking feature beyond the announcement, nor does it detail the technical mechanism OpenAI will use to embed or detect the marks. It also does not indicate whether the measure will extend to other OpenAI products or regions outside the EU.

The watermarking requirement sits alongside other provisions of the AI Act governing transparency for AI-generated content, as OpenAI continues adjusting its products to meet regulatory obligations in the bloc.