OpenAI Adds Text Watermarking to ChatGPT and Codex, Starting in the EU

OpenAI announced in October 2026 that it is rolling out text watermarking across ChatGPT and Codex, beginning with users in the European Union and opt-in access for API customers worldwide.

Over the coming weeks, text watermarking will be introduced for eligible ChatGPT and Codex users across all plans in the EU. OpenAI stated it is not making text watermarking a global default at launch, describing the regional approach as an opportunity to learn from real-world use and feedback.

Starting immediately, API customers around the world can opt in to watermarked text outputs for select models. OpenAI said this allows customers to determine how watermarking fits their transparency obligations and the experiences they provide to users. The company is also working with cloud partners to make watermarking available for OpenAI model outputs accessed through their services in the coming weeks.

Alongside the rollout, OpenAI is granting approved researchers and expert organizations access to a watermark detection tool, with applications open as of the announcement date. Access will initially be granted on a case-by-case basis to support evaluation and improvement of text provenance. The detector will report whether it identifies an OpenAI watermark, without revealing the user’s identity, prompts, or conversations. OpenAI noted that due to the risk of missed watermarks and false positives, the tool will not be made publicly available at launch.

The EU-first approach suggests the rollout may be connected to regional regulatory requirements, such as the EU AI Act, though OpenAI did not specify the exact compliance driver in its announcement. For readers and businesses that use AI-generated content, watermarking could provide a mechanism to verify whether text was produced by an OpenAI model — a capability that may carry implications for transparency and content authenticity.