OpenAI has announced that it will begin adding invisible watermarks to texts generated by ChatGPT and Codex under the EU Artificial Intelligence Act. EU transparency rules, which entered into force on August 2, require AI-generated content to be marked in a way that systems can recognize. The feature will be rolled out gradually over the coming weeks to eligible users on all plans and will be limited to the EU for now. Developers will be able to enable it on the models of their choice via the API starting today; the feature will be off by default. The company is not planning a global default. The watermark carries a pattern embedded in word choices that does not disappear when copied. In the “textGrain” report, it was stated that detectors can determine whether content is AI-generated using the text and a secret key.
Background
ChatGPT is not a new name in the FikirPilot archive: over the last 90 days we have published 37 stories featuring this name; the most recent is dated October 6, 2026.