OpenAI adds text watermarks for EU users
OpenAI began adding invisible, machine-readable watermarks to text produced by ChatGPT and Codex for users in the European Union, according to five technology outlets. The change was designed to meet EU AI Act requirements. OpenAI also offered global API customers an opt-in watermarking setting, while acknowledging editing could weaken detection.
Why it matters
The rollout tests whether text provenance rules can work at scale without making ordinary AI output visibly different.
Sources
finite. summarises the reporting above and links to each original. We do not reproduce full articles. Read the sources for complete coverage.
Part of the October 6, 2026 brief.