OpenAI’s new text watermark, textGrain, embeds a statistical pattern in word choices rather than adding hidden characters. It is designed to help detect some text generated or processed by OpenAI systems, but the company’s own tests show that detection can fall sharply when text is short, constrained, or lightly edited. The launch is also limited: ChatGPT and Codex watermarking is planned first for eligible users in the EU, while API customers worldwide can opt in for selected models.
What OpenAI announced—and who gets it
OpenAI announced textGrain on October 5, 2026. The company says API customers worldwide can opt in to watermark outputs from selected models. It plans to roll out watermarks on eligible ChatGPT and Codex text in the EU over the following weeks, across all plans. OpenAI said it was not making watermarking a global default at launch, and that the consumer-app rollout was EU-only at that point. It is also working with cloud partners to extend watermarking to eligible outputs served through them. OpenAI’s announcement
As an Amazon Associate I earn from qualifying purchases.
OpenAI links its EU rollout to machine-readable marking requirements under the EU AI Act and its commitments under the EU Code of Practice on Transparency of AI-Generated Content. That is the company’s explanation for the rollout; it does not settle every question about which content or providers are covered by law. OpenAI’s announcement
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsHow textGrain puts a mark in text
OpenAI says textGrain subtly shifts a model’s random choices among plausible words or word pieces. Across a passage, those choices create a statistical pattern a detector can search for. The mark is in the wording pattern: OpenAI says the system does not insert hidden characters, invisible spaces, unusual punctuation, or extra watermark-only tokens. This is different from a visible “AI-generated” label and from a classifier that examines text after generation without an embedded signal. OpenAI’s announcement
#1 Best Overall
API customers can enable the feature at organization or project level and choose which supported models receive a watermark. Eligibility can change, so customers should check their current settings rather than assume every model supports it. Turning on watermarking does not automatically grant access to the text detector. OpenAI Help Center
How well does OpenAI’s detector work?
The performance figures below are from OpenAI’s 2026 evaluation, not an independent audit. Detection rates describe test passages under particular conditions; they are not a guarantee for any individual text.
| OpenAI test condition | Company-reported result |
|---|---|
| 200-token psychology passages, with a target false-positive rate of 1% | About 80% detected |
| 400-token psychology passages, with a target false-positive rate of 1% | About 95% detected |
| 400-token passages after replacing 10% of words with synonyms | Detection fell from about 92% to 66% |
| 400-token passages after replacing 25% of words with synonyms | Detection fell to 17% |
OpenAI reports substantially lower detection on mathematics, where there is less room to vary word choice. Together, the results show why length, subject, and editing matter: a detector may recognize a watermark in a longer, flexible passage and miss it in a shorter, constrained, or rewritten one. The Register’s October 6 coverage called the approach “weak sauce,” a critical characterization of these limitations rather than a formal performance category. The Register’s report
OpenAI also says watermarking did not produce meaningful benchmark differences for its Astra model; its Help Center describes observed differences as within the noise of evaluation runs. Those are company-reported quality results. The available sources do not establish an independent measurement of textGrain’s quality impact in deployed use. OpenAI’s announcement OpenAI Help Center
Rank #3
- Used Book in Good Condition
What a positive or negative result can tell you
OpenAI says a positive detection can indicate that an OpenAI system generated or processed part of a passage. It cannot tell you how much human judgment, editing, or creativity contributed, identify a user or account, reveal a prompt or conversation, establish ownership or responsibility, or verify that the text is true, accurate, harmful, or presented in context. OpenAI’s announcement
A negative result is not proof that a person wrote the text. A watermark may be missed if the passage is short or constrained, has been edited or translated, came from an unsupported model, predates watermarking, or was produced with another provider’s tools. Detection is a limited provenance signal, not an authorship verdict.
Rank #4
Can the public check text for an OpenAI watermark?
Not through a public self-service text checker at launch. OpenAI says approved researchers and expert organizations may apply for detector access, initially considered case by case. Its Help Center lists research into provenance and detection reliability among possible qualifying uses. OpenAI’s announcement OpenAI Help Center
OpenAI’s image and audio provenance tools are a separate offering. The Help Center says those verification tools are publicly accessible to organizations, while text-detector access is restricted. OpenAI also says machine-readable provenance signals do not replace visible labels or other notices that may be required. OpenAI Help Center
What remains uncertain about independent verification
A September 2026 preprint by Alexander Nemecek, Vipin Chaudhary, and Erman Ayday argues that deployed watermark systems are difficult to verify without access and shared evaluation methods. Its experiments examine an open-source SynthID-Text implementation on two open-weight models—not OpenAI’s textGrain—so they do not validate or refute OpenAI’s reported figures. September 2026 preprint
OpenAI says textGrain matched or exceeded the approaches it tested, including SynthID for text, but that comparison is also the company’s claim. A separate independent evaluation of textGrain’s deployed performance is not established by the available sources. OpenAI itself cautions: “Text watermarking and detection remain early technologies with significant limitations, and views about their benefits and responsible uses are still developing.” OpenAI’s announcement
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →
