OpenAI will watermark ChatGPT outputs by default—but only in the EU

OpenAI has announced it will begin automatically watermarking text generated with ChatGPT in the European Union. The feature will also be offered in other regions, but will be off by default outside the EU.

The company says the European move is driven by the need to comply with the EU AI Act, which took effect in August and requires marking content produced by AI models in a way another tool can detect. OpenAI said the watermarking will roll out to EU users in the coming weeks.

OpenAI's method is proprietary and called textGrain, and the company has published a technical paper explaining how it works. In general, according to the report, it places patterns in word choices that are not clear to a human reader and do not meaningfully change the general quality of output, but that someone with a key can use a specialized detector to find. OpenAI says it will give detector access to a limited number of researchers and organizations, with a request-for-approval process for others to be added over time.

OpenAI's tests of textGrain show a 92 percent successful detection rate, but the tests also show that changing just 10 percent of the text in an output reduces the successful detection rate by almost 30 percent, and changing 20 percent of the text can lower the success rate by almost 75 percent. The success rate is generally lower for shorter or translated text than for longer text.

In August, OpenAI competitor Anthropic also introduced watermarking for text generated by its models, but Anthropic has enabled it globally, while OpenAI is currently only making it the default where regulators require it, at least in the ChatGPT and Codex apps. It will remain off but optionally available in the API, it seems.