# OpenAI will watermark ChatGPT text in the EU | QANATONE > OpenAI's textGrain method leaves a statistical trace in word choices. It comes first to ChatGPT and Codex users in the EU to comply with… Kaynak: https://www.qanatone.com/en/haber/openai-ab-metin-filigrani/ --- [News](https://www.qanatone.com/en/haber/) # OpenAI will watermark ChatGPT text in the EU 5 Oct 2026 3 min read News · Policy OpenAI's textGrain method leaves a statistical trace in word choices. It comes first to ChatGPT and Codex users in the EU to comply with the AI Act's transparency rules. ## What happened On October 5 OpenAI said it will add an invisible watermark to text produced by ChatGPT and Codex in the European Union. The system, called textGrain, is designed to meet the EU AI Act transparency rules that took effect on August 2, and rolls out over the coming weeks to eligible users on all plans in the EU. ## How it works textGrain adds no hidden characters. It gently shapes the model's word choices so that a statistical pattern sits in the words themselves. Readers cannot see it, but a detector holding a secret key can. Because the signal lives in the words, it survives copy and paste, and OpenAI says it does not identify the user. **92% → 66%** detection accuracy after 10% of words are swapped for synonyms ## Limits Rewriting erodes the signal. Swapping 10% of words for synonyms drops detection from about 92% to 66%, and heavier paraphrasing should push it lower. The watermark is a deterrent and a way to flag large-scale content farms, not proof of authorship. ## Why text is hardest Images offer thousands of pixels to hide a mark; in text every word carries meaning, and pushing word choice too hard hurts quality. textGrain balances this by nudging choices between near-synonyms. ## The API From October 5, API developers anywhere can switch the watermark on, but it stays off by default outside the EU ChatGPT rollout. ## What the AI Act requires The AI Act sets transparency duties for general-purpose models and the companies deploying them, starting with machine-readable marking of AI-generated or altered content. Images and audio can carry marks in metadata or pixels; text can only carry them in the word choices themselves, which is the approach textGrain takes. ## Why it matters The AI Act requires machine-readable marking of AI-generated content. Text has been the hardest medium to mark, and OpenAI's move sets a bar for other providers. ## What it means for your business Publishing raw AI text is becoming more traceable, at least in the EU. The durable answer is the same as for search visibility: make content original through expertise, your own data and real editing. ## More articles [News · Models ### Reflection unveils Beam, a 501B-parameter open-weight model 5 Oct 2026 · 3 min](https://www.qanatone.com/en/haber/reflection-beam-acik-agirlik-modeli/)[News · Ads ### TikTok launches an AI shopping assistant and one-click checkout 5 Oct 2026 · 3 min](https://www.qanatone.com/en/haber/tiktok-yapay-zeka-alisveris-asistani/)[News · Models ### Flaw found by Mythos exploited a day after disclosure 3 Oct 2026 · 3 min](https://www.qanatone.com/en/haber/mythos-rejetto-acigi-istismar/)[News · Infrastructure ### Apple tightens macOS Full Disk Access over AI agent risks 2 Oct 2026 · 3 min](https://www.qanatone.com/en/haber/apple-tam-disk-erisimi-ajanlar/)[News · Science ### Meta's Muse Spark co-authors six mathematics papers 2 Oct 2026 · 3 min](https://www.qanatone.com/en/haber/meta-muse-spark-matematik-makaleleri/)[News · Search ### Judge throws out two antitrust suits over Google AI Overviews 1 Oct 2026 · 3 min](https://www.qanatone.com/en/haber/ai-overviews-davalari-reddedildi/)[Data · Performance ### We tested award-winning agency sites: most fail on mobile 1 Oct 2026 · 4 min](https://www.qanatone.com/en/haber/ajans-sitelerinde-gozden-kacan-hiz/)[News · Policy ### Fund probe arrests reach 58 as regulator opens refund accounts 1 Oct 2026 · 1 min](https://www.qanatone.com/en/haber/fon-sorusturmasi-iade-hesabi-tutuklama/)