Skip to content
GrokDetect
EN

Grok AI detector facts

By Yang Zheng · Updated

Short answer: Grok's text watermark is off by default, nobody outside the vendor's approved applicants can check it, and GrokDetect works from writing style on Fileverity's engine. Every fact below carries a date and a source.

Watermark

Naming

Versions

Detector accuracy

These figures come from the accuracy page and cover Veread, Fileverity's text model:

Family results are writing-style similarity, never proof of authorship. GrokDetect does not read the vendor's watermark. For longer texts and reports, use the full AI text detector.

Questions and answers

Is Grok text watermarked by default?

No. The vendor documentation updated on 9 November 2026 says text watermarking is off by default and team admins can turn it on per model.

How accurate is the Grok detector?

On the accuracy page, 0.5% of human-written texts in seven languages were wrongly flagged, and 100% of fully AI-written xAI texts were detected.

What is the shortest text the detector accepts?

100 words.