OpenAI / TechCrunch / The Verge
Important
OpenAIPolicyEUWatermarkingOpenAI rolls out textGrain watermarks for ChatGPT and Codex text in the EU
October 5, 20262 min read
OpenAI will add invisible textGrain watermarks to eligible ChatGPT and Codex text output in the EU over coming weeks to meet EU AI Act provenance rules. API customers worldwide can opt in from October 5 (off by default). Detection is strong on longer passages but drops sharply if text is lightly edited.
Why it matters
This is OpenAI’s concrete response to EU machine-readable labeling rules — regional first, not global default. It also shows the practical limit of text watermarks: a modest synonym pass can collapse detection rates.
OpenAI published its approach to EU text provenance rules on October 5. Over the coming weeks it will add an invisible watermark to eligible ChatGPT and Codex text output for users in the European Union across all plans. API customers globally can opt in to watermarking for select models starting the same day; the feature remains off by default in the API. OpenAI said it is not making text watermarking a global default at launch.
The system, textGrain, embeds a statistical signal in the model’s word choices so a detector can assess whether a passage carries an OpenAI watermark. OpenAI reported roughly 80% detection on 200-token psychology passages and about 95% on 400-token passages at a 1% false-positive target, with substantially lower performance on math. Replacing 10% of words with synonyms cut detection on 400-token passages from about 92% to 66%; replacing 25% cut it to 17%. Detector access is initially limited to approved researchers and expert organizations. OpenAI said the watermark does not identify the user and showed no meaningful performance hit on listed Astra benchmarks.