OpenAI enables ChatGPT text watermarking by default in the European Union

OpenAI will begin automatically watermarking text generated by ChatGPT by default in the European Union, Ars Technica and The Verge reported. Ars Technica added that OpenAI will also offer the text watermarking feature in other global regions, but it will be turned off by default outside of the European Union. [1][3]

According to Ars Technica, the move in Europe is driven by the need to comply with the EU AI Act, which came into effect in August. The law mandates that content produced by artificial intelligence models must be marked in a manner that allows another software tool to detect it. [1]

Ars Technica reported that current watermarking solutions are not especially reliable and remain easy to circumvent. Existing technical standards like SynthID and the C2PA project can be bypassed, and there is still no completely effective method for marking generated text. [1]

The company is also testing a new visual ad format in ChatGPT, according to reports from The Information and Search Engine Journal. The Information reported that the visual ads will use images to display products and services while users generate images, with testing starting later this month. [2][6]

OpenAI reached a valuation of approximately $850 billion after major funding rounds in early 2026. As of September 2026, ChatGPT has grown to over 1.2 billion weekly active users across global markets. [8][9]