Add as a preferred source on GooglexAI Will Add Invisible Watermarks to Grok API Text Nov. 7
The default setting covers five model versions across text-generation API endpoints. Team administrators can disable it per model for specific legal exceptions and compliance needs.
xAI will add invisible statistical watermarks by default to text from five Grok API models starting Saturday, Nov. 7, giving generated output a detectable pattern without displaying a label.
The change covers grok-4.7, grok-4.6, grok-4.5, grok-4.3 and grok-4.20. It will apply across text-generation endpoints, including api.x.ai, the Responses API and the Chat Completions API.
The watermark subtly steers the model’s token choices toward a pattern. The text has no visible marker, but the statistical signal can emerge when enough output is analyzed.
Team administrators can turn off watermarking for individual models through the xAI Console. The opt-out is limited to specific legal exceptions and compliance needs.
The update applies to the API and does not affect consumer-facing Grok products. API users without an eligible opt-out will receive watermarked text by default starting Nov. 7.
Simon Yoon
ReporterSimon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.