xAI to Watermark Grok Text by Default

Starting November 7, 2026, invisible watermarking will be enabled across Grok API text models unless administrators opt out.

Updated on Oct. 31, 2026 in Artificial Intelligence

Isometric editorial illustration of matte storage cylinders in an industrial hall, with a thread of dotted light passing through a translucent filter plate. — xAI will begin invisibly watermarking Grok API text output by default on November 7, 2026, reversing its earlier position against altering AI-generated text. AI Illustration. Upload story photo >

Live Poll

Do you trust that AI-written text is clearly identified when you read it online?

xAI will switch on invisible statistical watermarking across its Grok API text models by default beginning November 7, 2026. The watermark covers several Grok model versions and applies to the api.x.ai endpoint, the Responses API, and the Chat Completions API.

Why it matters

The change marks a shift in xAI's stated position, since the company had previously argued that AI text output should remain unaltered. The update aligns with the transparency demands of the EU AI Act, which requires greater transparency from generative AI systems.

The watermark nudges token selection toward a statistical pattern while leaving visible text unaltered across five Grok model versions and three endpoints. Detection reliability and resistance to editing were not addressed in the announcement.

The players

xAI

xAI is the artificial intelligence company founded by Elon Musk that develops the Grok family of large language models.

The details

Team administrators can disable the watermark for each model through the xAI Console, and consumer-facing products are not affected by the update. The change is being rolled out on the API side only, with activation set for November 7, 2026.

October 31, 2026: xAI announces the Grok text watermarking change on X.
November 7, 2026: Watermarking switches on by default across Grok API models.

Regulatory Pressure

The EU AI Act requires greater transparency from generative AI systems, and xAI's default watermarking on Grok API text models reflects that regulatory direction. The change also reverses the company's previous argument that AI text output should remain unaltered.

Developers using the xAI API will see watermarked text by default from November 7, 2026, unless they disable it per model in the xAI Console. Consumer-facing Grok products are not affected by the change.

The takeaway

Developers building on Grok should review their API settings before November 7, 2026, to decide whether to keep the watermark enabled or disable it for each model. Teams that depend on unaltered model output should test their pipelines against the watermarked endpoints in advance.