Breaking

OpenAI to Add Invisible textGrain Watermarks to ChatGPT and Codex Text in the EU

October 5, 2026 at 20:11 EDT

Policy & Regulation
Foundation Models
Software Dev & Coding

OpenAI said it will add invisible textGrain watermarks to eligible ChatGPT and Codex text output in the European Union over the coming weeks, making its generated text machine-readable under the EU AI Act. The company is also letting API customers worldwide opt in for select models while keeping API text watermarking off by default at launch.

Oct. 5, 2026 · OpenAI

Invisible text watermarks enter ChatGPT

OpenAI will add textGrain watermarks to eligible ChatGPT and Codex text in the EU, while API customers can opt in globally for select models.

EU rollout

Coming weeks

False-positive target

400-token detection

95%

25% synonym edit

17%

Editing weakens the signal

In 400-token passages, synonym replacement sharply reduced detection. Each block represents about five percentage points.

How the rollout works

Eligible EU ChatGPT and Codex output

textGrain steers word choices

Approved detector checks the signal

What it can help show

A passage may contain a machine-readable signal associated with OpenAI-generated or processed text.

What it cannot prove

It does not identify a user, prove authorship, verify accuracy, or survive every rewrite, translation or short passage.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.