OpenAI said Monday that it will add an invisible watermark to eligible text output from ChatGPT and Codex to comply with the European Union's AI Act in the coming weeks.

Under the act, AI providers must add marks to enable the detection of AI-generated or manipulated content, the European Commission said earlier this year.

OpenAI said developers using its API can opt in to text watermarking for select models, but noted it will remain off by default.

The Microsoft-backed (MSFT) company said it will initially provide access to a limited number of approved researchers and expert organizations.