Summary: Elon Musk failed to appear for a voluntary interview with Paris prosecutors investigating Grok’s generation of an estimated 23,000 sexualised images of children and 3 million sexualised images overall over an 11-day period, as the US DOJ refused to assist the French probe. The case, which covers five suspected criminal offences including complicity in child pornography, is one of more than a dozen international legal actions against xAI, while Paris prosecutors have separately alleged the deepfake crisis may have been orchestrated to boost the value of the SpaceX-xAI entity ahead of its planned $1.75 trillion IPO.
Elon Musk failed to appear today for a voluntary interview with Paris prosecutors investigating Grok’s generation of sexualised images, including an estimated 23,000 images appearing to depict children over an 11-day period in late December and early January. The Paris prosecutor’s office told AFP it had “taken note” of the absence of those summoned. Linda Yaccarino, the former CEO of X, was also called to testify. Other X employees are scheduled to be heard as witnesses throughout this week.
Weeks before the summons, Musk dubbed French authorities “retards” in a French-language post on X. He previously called the February police raid on X’s Paris offices a “political attack.” The US Department of Justice refused to assist the French investigation on 18 April, telling French law enforcement in a letter that “this investigation seeks to use the criminal legal system in France to regulate a public square for the free expression of ideas” and that it was contrary to the First Amendment. The Paris prosecutor’s office responded that “the French constitution guarantees the separation of powers and the independence of the judiciary.”
What the investigation covers
The case, led by the cybercrime unit of the Paris prosecutor’s office, was opened in January 2025 after complaints alleging that X’s algorithms were biased and used to interfere in French politics. It expanded in November 2025 to include five suspected criminal offences: complicity in possessing and spreading pornographic images of minors, distribution of sexually explicit deepfakes, denial of crimes against humanity, manipulation of an automated data processing system as part of an organised group, and fraudulent data extraction.
The deepfake charges centre on Grok’s image generation capabilities, which allowed users to upload photographs of real women and girls and receive sexualised or nude versions without the subjects’ consent. The Center for Countering Digital Hate estimated that Grok generated approximately 3 million sexualised images between 29 December 2025 and 8 January 2026, including roughly 23,000 that appeared to depict children. The rate peaked at 190 sexualised images per minute, or one child image every 41 seconds. Up to 41% of the 4.6 million total images Grok produced during that period contained sexual imagery of women, according to data cited in class action filings.
The image generation feature had a troubled history before the crisis. xAI launched its Aurora model on 9 December 2024 but pulled it within hours after it generated photorealistic images of real people without safeguards. Grok’s “spicy mode” generated explicit content by design. When Musk announced on 20 December 2025 that Grok could edit and generate images directly on X, abuse exploded. On 9 January 2026, xAI restricted image generation to paid subscribers. On 14 January, it said it had blocked nudification capabilities entirely. But retests by NBC News in February showed Grok was still producing sexualised images, and in March the Dutch organisation Offlimits demonstrated it could still generate a sexualised video of a real person from a single uploaded photograph.
The global response
France is not acting alone. Malaysia and Indonesia became the first countries to block Grok entirely on 11 and 12 January. Japan’s Cabinet Office summoned X Corp’s Japanese subsidiary. The European Commission launched a formal investigation into X under the Digital Services Act in late January, ordering X and xAI to preserve all internal documents and technical data related to Grok until the end of 2026. The Amsterdam District Court ordered xAI to stop generating non-consensual nude images in the Netherlands on 26 March, with fines of EUR 100,000 per day for noncompliance. The UK’s Information Commissioner and Ofcom both opened investigations. Switzerland’s finance minister, Karin Keller-Sutter, filed criminal charges after Grok generated misogynistic abuse about her on X, the first time a serving head of a national finance ministry pursued criminal action against AI-generated content.
In the United States, the Senate unanimously passed the DEFIANCE Act on 13 January, creating a federal civil cause of action allowing victims to sue for $150,000 to $250,000 per violation, though it still awaits House approval. California’s attorney general launched an investigation and issued a cease and desist. Baltimore became the first US city to sue xAI. Three Tennessee teenagers filed a class action on 14 April alleging Grok generated pornographic deepfakes from their real photos, with the images spreading to Discord, Telegram, and the dark web. The lead class action, Jane Doe v. xAI Corp., was filed in the Northern District of California on 23 January.
The financial dimension
Paris prosecutors alerted the US DOJ and SEC in March, suggesting that the deepfake controversy “may have been deliberately orchestrated to artificially boost the value of the companies X and xAI” at a time when X was “clearly losing momentum.” SpaceX acquired xAI in a $1.25 trillion merger on 2 February 2026. The combined entity filed a confidential draft registration statement with the SEC on 1 April for a June 2026 Nasdaq IPO targeting a $1.5 to $1.75 trillion valuation, with 21 banks enlisted to raise $50 to $75 billion.
xAI’s standalone financials before the merger showed $107 million in quarterly revenue and a $1.46 billion net loss. Grok has roughly 60 to 64 million monthly active users. A quarter of European organisations have banned Grok, compared with roughly 10% for ChatGPT and Gemini.
Musk’s pattern of defiance
The decision not to appear in Paris is consistent with Musk’s broader posture toward European regulators. He has publicly considered pulling X out of the EU to avoid Digital Services Act compliance. X was previously found to be training Grok on user data without proper notification, likely breaching EU data protection rules. The Stanford AI Index, published this week, gave xAI a transparency score of 14 out of 100, among the lowest of any frontier model developer.
The DOJ’s refusal to cooperate with French prosecutors effectively shields Musk from the investigation’s most immediate consequences. A voluntary interview carries no penalty for nonappearance under French law. But the investigation remains active, the charges are serious, and France can issue a European arrest warrant if prosecutors escalate the case. Musk cannot visit France or any EU country that would enforce such a warrant without risk. For a man who has built a $1.25 trillion entity on the premise that rules are obstacles to be overcome, the Paris prosecutor’s office is presenting a test of whether that premise applies when the rules involve the sexual exploitation of children.
