OpenAI will watermark ChatGPT text in the EU but makes it optional for API users worldwide
OpenAI is rolling out an invisible watermark called textGrain for ChatGPT and Codex text in the European Union to comply with the EU AI Act, which took effect on August 2, 2026. The watermark will be enabled by default for EU users over the coming weeks, while API users worldwide can opt in starting immediately, though it remains off by default. The system embeds a statistical signal in word…
Key points
- OpenAI will enable textGrain watermarking by default for ChatGPT and Codex in the EU over the coming weeks.
- API users worldwide can opt in to watermarking starting today, but it remains off by default.
- At 1% false-positive rate, detection is ~95% for 400-token psychology passages and ~80% for 200-token; drops to ~66% and ~17% with 10% and 25% synonym replacement.
Sources agree on the core mechanics: textGrain works like SynthID, detection rates are about 95% for 400-token psychology passages and 80% for 200-token ones at a 1% false-positive rate, and editing significantly reduces detection — replacing 10% of words drops it to ~66%, 25% to ~17%. Detection is lower for math due to limited word flexibility. Access to the detector is initially restricted to approved researchers and expert organizations under the EU’s Code of Practice.
Minor differences appear in framing: TechCrunch notes the watermark rolls out to eligible users on all plans in the EU, while SearchEngineJournal adds that OpenAI previously held back watermarking due to fears users would switch to rivals, per a 2024 Wall Street Journal report. IBTimes emphasizes the technology’s early-stage limitations and OpenAI’s intent to revisit its approach as standards evolve.
The story so far
6 episodes →- OpenAI will watermark ChatGPT text in the EU but makes it optional for API users worldwidethis story
OpenAI will watermark ChatGPT text in the EU but makes it optional for API users worldwide
The Decoder · 5 October 2026
Loading the full article…
This text was published by The Decoder and written by Matthias Bastian. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
10sources- OpenAI Adds Invisible ChatGPT Watermark in EU Under AI ActPress · coincentral.com ·
- OpenAI to watermark ChatGPT text in Europe under AI Act: How it will workPress · economictimes.indiatimes.com ·
- OpenAI To Watermark ChatGPT Text In The EU, Opens API Opt-In via @sejournal, @MattGSouthernPress · Search Engine Journal ·
- Big Update For ChatGPT Users: Your AI-Generated Text Could Soon Carry A Hidden Watermark; Here's WhyPress · goodreturns.in ·
- OpenAI will start watermarking ChatGPT’s text in the EUPress · TechCrunch AI ·
- ChatGPT Gets an Invisible Watermark in Europe: What OpenAI's New AI Rule MeansPress · ibtimes.sg ·
- OpenAI is adding text watermarking in ChatGPT and CodexPress · The Verge AI ·
- Secret stamp: OpenAI adds invisible watermarks to ChatGPT text under EU rulesPress · interestingengineering.com ·
- An 'Invisible Signature' on My Writing! ChatGPT's Text Watermark 'textGrain' Starts in the EU | Hyper House Morning AI News (10/6)Press · note.com ·
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- Trump urges states to welcome AI data centers after ballot defeats · 1 src
- AI teddy makes unprompted 'I can't help with killing you' remark · 1 src
- OpenAI engineer charged with planting bug in HQ denies spying · 1 src
- Goldman Sachs restricts AI coding tools for non-technical staff amid security risks · 3 src
- Congresswoman Ansari calls for artificial superintelligence ban after midterm sweep · 2 src
Comments
via GitHub Discussions