# textGrain Explained

> OpenAI's textGrain is an invisible watermark for ChatGPT and Codex text. This page explains what it is, who it affects, how accurate detection is, and why no public checker exists.

textGrain went live on October 5, 2026. OpenAI nudges the words its models pick, so a detector holding the secret key can later test whether a passage carries the pattern. No hidden characters. Nothing visible. Copy-paste does not wash it out.

The EU gets it first. Eligible ChatGPT and Codex output there is watermarked on every plan, phased in over weeks, with no opt-out. That is the EU AI Act's transparency rule at work, in force since August 2, 2026. API developers anywhere can opt in for select models. It stays off by default. OpenAI is not making this a global default.

On accuracy: at a 1% false-positive target, the detector caught about 80% of 200-token passages and 95% of 400-token passages, on ordinary prose. Math and short snippets scored much worse. Editing weakens the signal fast. In OpenAI's tests, replacing 10% of words with synonyms dropped detection from 92% to 66%. Replacing 25% dropped it to 17%.

A positive result means an OpenAI system likely touched the passage. It does not name the author. It does not measure how much a person edited. A negative result carries no meaning on its own.

You cannot check for it publicly. OpenAI gives the detector only to approved researchers and expert organizations, case by case. Its public verify tool covers images and audio, never text. Anyone selling a "ChatGPT watermark detector" is guessing. They do not have the key.

For comparison: Anthropic watermarks every Claude output worldwide with no opt-out, and users complained in August 2026. Google DeepMind's SynthID-Text is the closest technical cousin. OpenAI sits in the middle.

This page is for anyone who heard ChatGPT is getting an invisible watermark and wants it explained in plain language. It is free. No account, no signup. An informational page; nothing runs in the browser and nothing is collected.

Guide: https://lab.awaara.cn/textgrain/
Related: LLMS.txt Checker https://lab.awaara.cn/llms-txt-checker/
Related: Claude Skill Validator https://lab.awaara.cn/claude-skill-validator/
Privacy: https://lab.awaara.cn/privacy/
Terms: https://lab.awaara.cn/terms/
All tools: https://lab.awaara.cn/