The EU AI Act requires generative AI providers to make generated text identifiable in a machine-readable way. OpenAI has already made tools publicly available to identify images and audio generated by its models, and now the company has shared its approach to text watermarking in response to the EU AI Act.
We're expanding our approach to content provenance to include text in response to EU regulatory requirements, while recognizing the significant limitations of current text watermarking technology.
Our tools already help verify whether an image or audio file was created with our…
— OpenAI (@OpenAI) October 5, 2026
Text Watermarking
OpenAI introduces its text watermarking technology, textGrain, that adds an invisible statistical signal to the model’s word choices. OpenAI’s detector looks for that signal to assess whether a passage contains an OpenAI watermark. The company also plans to make the technology available in open source so that others can build on it.
During evaluations, textGrain matched or exceeded the performance of other approaches that were tested, including SynthID for text. But strong performance under ideal conditions does not guarantee reliable detection in everyday use.
Limitations of text watermarking
OpenAI has also mentioned a few limitations of text watermarking: Shorter or more constrained text is harder to detect, and editing can weaken the watermark. Also, text watermarking provides a limited signal about the role OpenAI’s systems played in a passage:
- A watermark does not measure human contribution
- A watermark does not establish ownership or responsibility
- A watermark does not identify the user
- A watermark does not verify accuracy
- The absence of a detected watermark does not prove human authorship
Availability
Over the coming weeks, text watermarking will be introduced to eligible ChatGPT and Codex users across all plans in the EU only. OpenAI is not making text watermarking a global default at launch.
API customers around the world can opt in to watermarked text outputs for select models. This lets customers decide how watermarking fits their transparency obligations and the experience 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.
Also, approved researchers and expert organisations can apply for detector access. In accordance with the Code of Practice, access will be initially granted on a case-by-case basis to support evaluation and improvement of text provenance. The tool will report whether it detects an OpenAI watermark, without identifying the user or revealing their prompts or conversations. Given the risk of missed watermarks and false positives, the company is not making it publicly available at launch.
Apart from text watermarking, OpenAI adds Content Credentials to supported image outputs and makes image and audio verification available through openai.com/verify and its Content Provenance API.
