textGrain is the name of OpenAI's text watermarking method, announced on October 5, 2026. It slightly changes how the model makes its random choices between possible words, using a secret key. The result is a statistical pattern that readers cannot see but a detector with the same key can test for. OpenAI plans to release it as open source.

Is ChatGPT text watermarked in the United States?

Not by default. OpenAI says the watermark is being introduced for eligible ChatGPT and Codex users in the EU only and is not a global default at launch. Text from an app built on OpenAI's API can carry the watermark anywhere, but only if the developer has switched it on in the API settings.

Does copying and pasting remove the ChatGPT watermark?

No. The watermark is part of the wording, so it stays when text is copied and pasted. Editing is different. In OpenAI's tests on 400-token passages, replacing 10% of words with synonyms cut detection from about 92% to 66%, and replacing 25% cut it to 17%. Translation and heavy paraphrasing can make it undetectable.

Does the ChatGPT watermark add hidden characters to my text?

No. OpenAI says the watermark does not insert hidden characters, invisible spaces, unusual punctuation or extra tokens. It only changes the statistical pattern of word choices while the text is being generated. That is why you cannot find it by searching a document for strange symbols, and why stripping invisible characters from a text has no effect on it.

Can a teacher check an essay for the ChatGPT watermark?

Not at the moment. OpenAI's text detector is open only to approved researchers and expert organizations, and OpenAI says academic integrity uses, such as checking student essays or theses, are outside the program for now. A result would not show how much of the work a student wrote, and a missing watermark does not prove a human wrote it.

Does the ChatGPT watermark reveal who generated the text?

No. OpenAI says the watermark does not identify the user and is not connected to a person, an account, a prompt or a conversation. Its detector only reports whether an OpenAI watermark was found. It cannot show how much a human contributed, or whether the model wrote the text or only edited something a user supplied.

AI Policy & Safety

Does ChatGPT Watermark Text? What Changed in October 2026

By The Agent5 founder·October 12, 2026·How we fact-check

OpenAI is adding an invisible text watermark to ChatGPT and Codex for users in the European Union, to comply with the EU AI Act. Here is how it works, who gets it, how reliable detection is and how Claude and Gemini compare.

ShareX LinkedIn WhatsApp

Key takeaways

OpenAI announced on October 5, 2026 that eligible ChatGPT and Codex users in the EU will get an invisible text watermark over the following weeks, while it stays off by default everywhere else.
The method, textGrain, steers the model's random word choices with a secret key and adds no hidden characters and no user identity to the text.
In OpenAI's own tests on 400-token passages, replacing 10% of words with synonyms cut detection from about 92% to 66%, and replacing 25% cut it to 17%.
There is no public detector for ChatGPT text: access is limited to approved researchers and expert organizations, and checking student work is outside the program.
EU AI Act Article 50 has applied since August 2, 2026, and AI systems that were already on the market must meet its marking rule by December 2, 2026.

Does ChatGPT watermark text? Since October 2026, it depends on where you live. On October 5, 2026, OpenAI said it will add an invisible watermark to ChatGPT and Codex text for users in the European Union, rolling out over the following weeks. Elsewhere it stays off unless a developer turns it on. This guide explains what the ChatGPT watermark is, who gets it, how it works, whether it can be detected or removed, and how Claude and Gemini compare.

What does it mean for students, writers and developers?

In short, no public tool can check for the watermark, and a result would not prove who wrote a text. Here is what that means for each group.

Students. A school or university cannot use OpenAI's text detector to check coursework for now, because academic integrity uses are outside its access program. A detection result would also not show how much of an essay a student wrote. Anthropic takes a different line and lists educational organizations among those eligible for its Claude detection preview.

Writers and publishers. A watermark does not settle who owns a text. Proofreading is a special case: Anthropic notes that when a model only corrects your writing, nearly all the words stay yours, leaving little to watermark. Article 50 also has a rule for deployers, meaning people or organizations that use an AI system in a professional role. If they publish AI-generated text to inform the public on matters of public interest, they must disclose that, unless the text went through human review or editorial control and someone holds editorial responsibility. Purely personal, non-professional use is outside the AI Act's scope.

Developers. In the API the watermark is off until you enable it. The organization default is under Organization settings, Data controls, Text provenance. A project can override it under Project Settings, Text provenance. There you turn on Allow text watermarking and pick the models. Enabling it does not give you detector access.

A safety note: Ars Technica reported on September 17, 2026 on a study by a researcher at Lasso Security. In it, Google's open-source SynthID-Text changed how six open models responded to harmful requests, especially under prompt injection, where crafted text tries to override a model's instructions. The study did not test Claude's own setup or OpenAI's textGrain. The report's advice to developers: test how your models and AI agents (models that act through tools) behave with watermarking switched on.

Does ChatGPT Watermark Text? What Changed in October 2026

Does ChatGPT watermark text now?

How does the ChatGPT text watermark work?

What does EU AI Act Article 50 require?

Can you detect the ChatGPT watermark?

Can you remove the ChatGPT watermark?

Is there a Codex watermark on code?

What does it mean for students, writers and developers?

How do Claude and Gemini compare with ChatGPT?

What happens next

Sources