SpaceXSI will add an invisible watermark to Grok API text starting November 7

Which outputs are watermarked

Elon Musk’s AI company SpaceXSI (formerly xAI) has added an “Output Watermarking” page to its developer documentation, last updated on Oct. 31. Starting Nov. 7, 2026, text responses from supported Grok models on its API will carry an invisible watermark. It is on by default. The list covers grok-4.7, grok-4.6, grok-4.5, grok-4.3 and the reasoning, non-reasoning and multi-agent versions of grok-4.20.

The watermark applies to every text generation API, including the Responses API and the Chat Completions API. Ordinary text responses, structured outputs and tool call arguments are all covered. Developers don’t need to change their requests. An “x-watermark: true” header on the response shows that the output was generated with the watermark.

How it works

According to the documentation, the watermark is applied as the model generates text: it slightly shapes which tokens the model picks. As a result, it adds no characters, markup or hidden text to the response. Detection is statistical: SpaceXSI scores a piece of text against the watermark and reports it as watermarked, not watermarked or uncertain. Longer text gives a stronger signal. Access to the detection API is granted by application. The company says the watermark does not affect quality or speed; it has not said how well it holds up against rewriting.

EU law, global scope

SpaceXSI says it watermarks text output to comply with the EU AI Act. According to Search Engine Journal, the transparency rules in Article 50 of the act began applying on Aug. 2, 2026. Even so, the watermark is enabled on every SpaceXSI API endpoint, including the global endpoint, api.x.ai, not only in the EU.

Team admins can turn it off per model in the SpaceXSI Console, and the setting reaches every API key on the team within five minutes. Opting out requires confirming that an exception under applicable law applies to the team’s use. The team also becomes solely responsible for any marking, disclosure or other legal requirement. Some models may be locked on, and every opt-in and opt-out is recorded in the team’s audit log.

Earlier stance and rivals

On Aug. 12, City AM reported that xAI had not signed the EU’s code of practice on transparency for AI-generated content, though it still had to comply with the AI Act. The same report said Anthropic had announced that new Claude models would embed an invisible watermark in generated text. On Oct. 5, OpenAI said it would add a watermark called textGrain to ChatGPT and Codex text in the EU; according to BleepingComputer, it stays off by default in the API. SpaceXSI’s page covers only the API and does not mention the Grok app.