OpenAI Launches Invisible Text Watermarking for ChatGPT and Codex in EU to Meet AI Act Requirements

OpenAI has begun rolling out invisible text watermarking through its new textGrain system to comply with the European Union's artificial intelligence transparency mandate, marking a significant step in implementing the EU AI Act's content provenance requirements. Starting October 5, 2026, API customers worldwide can opt into the watermarking feature—disabled by default—while ChatGPT and Codex users in the European Union will receive automatic watermarking on all plans within the coming weeks.