Grok API answers get an invisible watermark Nov 7: what developers need to know
Quick answer
Starting Saturday, Nov 7, 2026, text generated through the Grok API carries an invisible watermark by default, according to xAI’s developer docs (xAI docs, updated Oct 31). You don’t need to change any code. To see whether a response was marked, read the x-watermark header: true means it was. Team admins can switch it off per model in the console, but only after confirming a legal exception applies.
Which models and outputs are covered
- Models: grok-4.7, grok-4.6, grok-4.5, grok-4.3, grok-4.20-reasoning, grok-4.20-non-reasoning and grok-4.20-multi-agent (xAI docs).
- APIs: every text generation API, including the Responses API and the Chat Completions API, on the global endpoint (api.x.ai) and on regional endpoints.
- What gets marked: ordinary text responses, structured outputs and tool call arguments.
- Why: xAI says it watermarks text output to comply with the EU AI Act.
How the watermark works
- It’s invisible. It slightly shapes which tokens the model picks, so it adds no characters, markup or hidden text to the response (xAI docs).
- xAI says it doesn’t change response quality or speed.
- Detection is statistical. xAI scores text as watermarked, not watermarked or uncertain, and longer text gives a stronger signal.
- The detection API isn’t public. You have to apply for access through a form.
What developers should do before Nov 7
- Nothing, if you’re fine with the default. There’s no request parameter, and the response format stays the same apart from the new header.
- Log the
x-watermarkheader if you need a record of which outputs were marked. - Decide whether to opt out. A team admin opens Team settings, finds the Model output watermarking row, clears the checkbox for each model, selects Save, then confirms with Opt Out. Changes apply to new requests within 5 minutes (xAI docs).
- Know what opting out means. You confirm that a legal exception applies to your use and that your team is responsible for any marking or disclosure the law requires. Every opt-in and opt-out is recorded in the team audit log.
FAQ
- Does opting out apply to all my API keys? Yes. The setting belongs to the team, applies to every API key on it, and is chosen separately for each model.
- Why can’t I turn it off for a model? That model is locked on for your team. Ask your team admin, then contact xAI support.
- Does this affect the Grok app or grok.com? xAI’s docs describe the API only. TokenPost reports the update “does not affect consumer-facing Grok products” (TokenPost).
- Will my users see anything different? No. The text looks the same; only the response header changes.