Elon Musk's Grok is the only major AI chatbot that hasn't signed the EU's watermarking code

Anthropic, OpenAI, Google, Microsoft, Meta, and Mistral have signed the EU's voluntary Code of Practice on Transparency of AI-generated content. Grok, the chatbot from Elon Musk's AI company xAI, is the only major AI lab that has not.

The code requires signatories to embed a watermark, an invisible, machine-readable signal, in AI-generated text, images, and audio so downstream tools and platforms can flag it as machine-made. Anthropic said on Monday it will place those signals in its content "from day one." OpenAI plans to watermark text outputs as part of the same commitment, saying it aims to expand provenance signals to all modalities including text. Google is expected to introduce measures for Gemini, though the specifics have not been made public.

The voluntary code sits under the EU AI Act, the bloc's binding law on artificial intelligence. xAI is still subject to the act's applicable requirements even though it did not sign the voluntary code, and may be compelled to introduce its own marking system regardless.

What remains unanswered is whether xAI will mark Grok's outputs at all, and whether EU regulators will treat the absence of a voluntary signature as a compliance signal in itself. The code is voluntary. The act is not.