Science & Technology

OpenAI introduces invisible watermark for ChatGPT texts in the EU

OpenAI will implement an invisible watermark for ChatGPT and Codex texts for users in the European Union, aimed at complying with AI legislation.

6 October 2026Tagblick Newsdesk

AI-generatedWritten by an AI system, editorially responsible — how we work

Archive photo: OpenAI introduces invisible watermark for ChatGPT texts in the EU — Archive photoPhoto: Eelens.paul at Dutch Wikipedia · Wikimedia Commons · CC BY-SA 3.0

OpenAI has announced that it will add an invisible watermark called textGrain to texts generated by ChatGPT and Codex. This will initially apply only to users in the European Union. The implementation of this feature is intended to meet the requirements of AI legislation within the EU.

The watermark is designed to help identify texts produced by artificial intelligence. It is a statistical signal embedded in the text, making it invisible to most users. OpenAI claims that the effectiveness of this watermark can reach up to 95 percent, but the detection rate drops significantly to 17 percent when the text undergoes many alterations.

In addition to OpenAI, Anthropic has also announced plans to introduce similar technology, but OpenAI has stated it will offer its API customers the option to disable the watermark. This could be a significant choice for developers providing AI-related services.

The watermark feature is a crucial step in enhancing trust in AI-generated content and raising user awareness about the origins of texts. The functionality will soon be rolled out in the EU, and it should make it easier for researchers and other stakeholders to verify whether a text has been generated by an AI system or not.

The technology used by OpenAI is claimed to perform better than other available techniques, including SynthID from Google DeepMind. This indicates that OpenAI aims to take a leading role in AI transparency.

Outside the EU, texts from ChatGPT and Codex will remain without watermarks. This creates a difference in how AI content is treated globally, emphasizing compliance with regional legislation.

This item was written by an AI system on the basis of the headlines and short teasers of the independently operating newsrooms named below. We disclose this under Art. 50(4), second subparagraph, of Regulation (EU) 2024/1689 (AI Act). Before publication we check origin and verifiability — anything that appears in only one place is not published — as well as language, substance and legally sensitive content. No separate substantive review of each individual item by a natural person takes place; we therefore do not rely on the exemption in Art. 50(4), second subparagraph, of the AI Act. Editorial responsibility for this publication rests with the person named in the imprint (§ 18 Abs. 2 MStV). Please report errors to [email protected] — we will correct or remove the item. Full reports are available from the sources listed.