The Paris public prosecutor's office announced Wednesday that it has opened a criminal investigation into X and Elon Musk on charges including complicity in possessing and distributing child sexual abuse images, unlawfully collecting personal data, and producing deepfakes. In a development that ties the case directly to the AI industry, prosecutors also cited Grok, X's AI system, for complicity in denying crimes against humanity through its outputs.

This is not a regulatory fine or a policy dispute. This is a criminal investigation by a G7 nation into the world's richest man and the platform he owns, with charges that carry prison time under French law.

The Charges

According to the AP, the investigation covers multiple categories of alleged criminal conduct. The most serious is complicity in possessing and distributing child sexual abuse material on the platform. French prosecutors allege that X failed to adequately remove CSAM despite being notified, making the platform complicit under French law.

The second category involves sexually explicit deepfakes distributed through the platform. The third involves unlawful collection of personal data. And then there is the Grok component: French prosecutors are investigating whether X's AI chatbot produced outputs that constitute denial of crimes against humanity, which is a criminal offense in France.

That last charge is the one the AI industry should be paying attention to. A government is treating AI-generated disinformation as criminal conduct, not by the user who prompted it, but by the company that built and deployed the AI system.

Timing and Context

The charges land at one of the worst possible moments for Musk. He is currently in the middle of his federal trial against OpenAI in Oakland. Grok usage has been declining throughout 2026 according to multiple reports. Every xAI co-founder has departed the company. And now a major European ally of the United States is opening criminal proceedings against him personally.

France's investigation also comes as the EU finalizes amendments to its AI Act. Last week, EU lawmakers agreed to water down high-risk AI regulations while simultaneously adding a new ban on AI-generated sexual deepfakes, a provision widely seen as a direct response to Grok. The timing of the French criminal charges suggests coordination between French domestic enforcement and EU-level regulatory signaling.

The Precedent Problem

If French prosecutors succeed in establishing that Grok's outputs constitute criminal conduct attributable to the company, the precedent extends far beyond X. Every AI company that deploys a chatbot or agent accessible in France would face potential criminal liability for what their models say. OpenAI, Anthropic, Google, Meta, and every other company running public-facing AI systems would need to assess their exposure under French criminal law.

The CSAM and deepfake charges are more familiar territory for platform liability law. But the crimes-against-humanity denial charge applied to an AI system is, as far as we can determine, a first. It treats an AI model's output as the legal equivalent of a human publication, with the deploying company liable as publisher.

X has not publicly responded to the charges as of Thursday morning. Musk has not posted about the investigation on his platform.

First reported by the Associated Press. Additional details from The Globe and Mail and multiple French media outlets.