OpenAI announced on October 5 that API customers worldwide can opt in to text watermarking for selected models. The setting remains off by default in the API.
Eligible ChatGPT and Codex text output in the European Union will receive an invisible watermark over the coming weeks. OpenAI said the launch responds to EU AI Act text provenance requirements.
Its textGrain system adds a statistical signal to word choices. Applications for detector access are open, but initial access is restricted to approved researchers and expert organizations.
Detection has practical limits
OpenAI says shorter passages and constrained subjects are harder to detect, while editing or translation can weaken the signal. A watermark does not verify accuracy, identify a user or measure human contribution. Likewise, a missing watermark does not prove human authorship. Public image and audio verification tools remain available.
Sources
OpenAI: Our approach to EU text provenance rules, October 5, 2026
Feature image: AI-generated editorial illustration for Guava Insights. Conceptual artwork, not a product photograph or depiction of an actual facility.
