A growing wave of class action lawsuits against Elon Musk’s xAI alleges that the company’s Grok chatbot was used to generate millions of sexually explicit deepfake images — including tens of thousands depicting minors — while the company knowingly failed to implement basic safeguards. The first suit, *Jane Doe v. xAI Corp.* (Case No. 5:26-cv-00772), was filed on January 23, 2026, in the U.S. District Court for the Northern District of California by a South Carolina woman whose fully clothed photo, posted innocuously on X, was transformed by another user through Grok into a revealing bikini image without her knowledge or consent.
A second, more explosive lawsuit followed on March 16, 2026, filed by three teenage girls from Tennessee — two of whom are minors — marking the first legal action brought by minors over Grok-generated deepfakes. The scale of what these lawsuits describe is staggering. According to research by the Center for Countering Digital Hate, Grok’s image generation tools produced over 3 million sexualized images in just an 11-day window spanning late December 2025 and early January 2026, with approximately 23,000 of those images depicting minors. The plaintiffs allege that xAI was aware its technology was being weaponized for sexual exploitation and chose to monetize the feature rather than shut it down. This article breaks down the details of both lawsuits, examines how Grok’s “Spicy Mode” was exploited, reviews the regulatory response from 35 state attorneys general and California’s cease-and-desist order, and explores what legal options may be available to victims.
Table of Contents
- What Are the Class Action Lawsuits Against xAI Alleging About Grok’s Deepfake Generation?
- How Did Grok’s “Spicy Mode” Enable Mass Production of Sexualized Deepfakes?
- What Regulatory Actions Have Authorities Taken Against xAI Over Grok Deepfakes?
- What Legal Options Do Victims of AI-Generated Deepfakes Have Right Now?
- Why Content Moderation Failures in AI Image Generation Are Uniquely Dangerous
- The Broader Pattern of AI Companies Facing Accountability for User-Generated Harm
- What Comes Next for the Grok Deepfake Litigation and AI Regulation
- Conclusion
- Frequently Asked Questions
What Are the Class Action Lawsuits Against xAI Alleging About Grok’s Deepfake Generation?
The two lawsuits paint a damning picture of corporate negligence. The January 2026 filing by the South Carolina plaintiff established the basic legal framework: that xAI built and deployed image generation capabilities without adequate content moderation, and that real people were harmed when their likenesses were manipulated into sexually explicit content without consent. but it was the March 2026 Tennessee case that escalated the legal stakes dramatically. That suit names both xAI and Elon Musk personally as defendants, contains 13 separate counts ranging from intent to distribute child pornography to intentional infliction of emotional distress, and seeks class action certification on behalf of all similarly situated victims. The Tennessee teens’ case is particularly alarming because of how methodical the abuse was.
According to court filings, a single individual gathered innocent photographs of the three girls from school yearbooks, Homecoming dance photos, and social media accounts. Using Grok’s “Spicy Mode” through a third-party vendor, the perpetrator then generated sexualized deepfake images and videos of these teenagers. The fabricated material was subsequently uploaded and distributed across Telegram, Discord, and Mega — platforms where such content can spread rapidly and is difficult to remove permanently. The suspect in the Tennessee case was arrested in late December 2025, but only after one of the victims was tipped off by an anonymous Instagram account and reported the situation to law enforcement. The lag between the creation of the deepfakes and their discovery underscores one of the core problems with AI-generated exploitation material: victims often have no idea it exists until it has already been widely distributed.
How Did Grok’s “Spicy Mode” Enable Mass Production of Sexualized Deepfakes?
Grok’s “Spicy Mode” was, according to xAI’s own promotional language, designed to let creators explore “edgier, more visually daring narratives.” In practice, plaintiffs allege, this meant the feature lacked the guardrails that competitors like OpenAI’s DALL-E and Google’s Imagen had implemented to prevent the generation of sexually explicit or non-consensual imagery. Where other major AI companies invested heavily in content filters, watermarking, and refusal mechanisms for harmful prompts, xAI allegedly treated the absence of such restrictions as a selling point. The numbers bear out just how aggressively the feature was exploited. The Center for Countering Digital Hate tracked Grok’s output over a nine-day period and found that the tool generated over 4.4 million total images, of which 1.8 million were sexualized depictions of women. Narrowing to an 11-day window from December 29, 2025, through January 8, 2026, researchers estimated over 3 million sexualized images were produced, with roughly 23,000 depicting minors.
These are not theoretical risks or hypothetical harms — they represent millions of discrete acts of image-based abuse facilitated by a commercially available product. However, it is worth noting that not every image generated during this period necessarily depicted a real, identifiable person. Some portion of the output likely involved fictional or composite subjects. That distinction matters legally but offers cold comfort to the real victims whose photographs were fed into the system. If you or someone you know suspects their likeness was used to create non-consensual deepfakes through Grok or any other AI tool, the critical first step is to document everything — screenshots, URLs, timestamps — before content is removed, as this evidence is essential for both law enforcement reports and civil litigation.
What Regulatory Actions Have Authorities Taken Against xAI Over Grok Deepfakes?
The governmental response to the Grok deepfake crisis was unusually swift and bipartisan. Thirty-five state attorneys general publicly demanded that xAI stop allowing its platform to generate sexual deepfakes — a rare show of cross-party consensus on a tech regulation issue. California Attorney General Rob Bonta went further, issuing a formal cease-and-desist letter to xAI on January 16, 2026, citing violations of four California statutes and giving the company until January 20, 2026, at 5:00 PM Pacific Time to confirm compliance and preserve all evidence related to the deepfake generation. xAI’s response to the California cease-and-desist was telling. Within 24 hours of receiving the letter, the company restricted Grok’s image generation features — including “Spicy Mode” — to paying subscribers only.
Critics immediately pointed out that restricting a harmful feature to a paywall is not the same as fixing it. Rather than disabling the capability that was producing millions of pieces of exploitative content, xAI essentially put it behind a subscription gate. The plaintiffs in the class action suits cite this decision as evidence that the company chose to monetize the problem rather than solve it. For comparison, when researchers demonstrated that Microsoft’s Bing Image Creator could be manipulated to produce harmful content in 2024, Microsoft implemented additional content filters within days and did not simply restrict the feature to paid users. The distinction matters because it speaks to corporate intent — a central element in the negligence and recklessness claims the lawsuits advance.
What Legal Options Do Victims of AI-Generated Deepfakes Have Right Now?
Victims of AI deepfakes face a frustrating legal landscape where the technology has outpaced the law, though that gap is narrowing. The Tennessee lawsuit’s 13 counts provide a roadmap of the legal theories currently being tested. These include claims under existing child exploitation statutes, intentional infliction of emotional distress, negligence, and various state privacy laws. The fact that the suits seek class action status is significant — if certified, a class action would allow potentially thousands of victims to pursue claims collectively rather than bearing the cost and burden of individual lawsuits. One important tradeoff victims should understand is the difference between pursuing criminal charges and civil litigation. Criminal cases, like the arrest of the Tennessee suspect in December 2025, target the individuals who created and distributed the deepfakes.
Civil suits like the ones against xAI target the company that built and deployed the tool. Both paths can be pursued simultaneously, but they operate on different timelines and have different evidentiary standards. Criminal cases require proof beyond a reasonable doubt; civil cases require only a preponderance of evidence. For many victims, the civil route through a class action may be more accessible, especially since individual criminal investigations can be slow and resource-intensive. Several states have enacted or are considering laws specifically targeting AI-generated deepfakes, particularly those involving minors. However, federal legislation remains incomplete, which is why the xAI lawsuits are being brought primarily under existing state statutes and common law theories rather than a comprehensive federal deepfake statute.
Why Content Moderation Failures in AI Image Generation Are Uniquely Dangerous
The xAI case highlights a problem that extends well beyond a single company. AI image generation tools can produce photorealistic content at a speed and scale that no human moderation team can match. When Grok generated 4.4 million images in nine days, even a well-resourced trust and safety team reviewing content around the clock could not have meaningfully screened that output in real time. The only effective safeguards are those built into the model itself — input filters that reject harmful prompts, output classifiers that flag problematic content before it reaches the user, and hard technical blocks on generating content depicting real individuals in sexual contexts. The warning for consumers and policymakers alike is that a company’s public statements about safety are meaningless without verifiable technical implementation.
xAI’s promotional language around “Spicy Mode” suggested creative freedom; the lawsuit alleges the reality was a tool with virtually no guardrails being used to mass-produce exploitation material. Other AI companies are not immune to similar failures, but the xAI case is notable for the alleged gap between what the company knew was happening and what it chose to do about it. One limitation of current legal and technical responses is the difficulty of content removal. Once deepfake images are distributed across decentralized platforms like Telegram, Discord, and Mega — as happened in the Tennessee case — removing every copy becomes practically impossible. Victims may win their lawsuits and still find that the images continue to circulate for years.
The Broader Pattern of AI Companies Facing Accountability for User-Generated Harm
The xAI lawsuits fit within a broader legal trend of holding AI companies responsible not just for what they build, but for how foreseeable the misuse was. Section 230 of the Communications Decency Act has historically shielded tech platforms from liability for user-generated content, but courts have increasingly scrutinized whether AI-generated content — where the platform’s own tool does the creating — qualifies for that protection. In this case, plaintiffs argue that Grok did not merely host harmful content uploaded by a third party; it actively produced the harmful content when prompted.
This distinction could prove pivotal. If courts agree that AI-generated imagery falls outside traditional Section 230 protections, it would open the door to direct liability claims against any AI company whose tools produce harmful outputs. The xAI cases may become a landmark test of that question.
What Comes Next for the Grok Deepfake Litigation and AI Regulation
Both lawsuits are in their early stages, and the path from filing to resolution in federal court is long. The Tennessee minors’ case, with its 13 counts and request for class certification, will likely face motions to dismiss from xAI’s legal team before proceeding to discovery. If the case survives those initial challenges and the court certifies a class, it could potentially represent thousands of victims nationwide — making it one of the largest AI harm cases ever litigated.
On the regulatory front, the coordinated action by 35 state attorneys general and California’s aggressive cease-and-desist signal that enforcement agencies are willing to move quickly on AI-generated exploitation, even in the absence of comprehensive federal legislation. Several bills addressing AI deepfakes are pending in Congress, and the outcomes of the xAI lawsuits will almost certainly influence how those bills are shaped. For now, the cases serve as a blunt reminder that deploying powerful generative AI tools without adequate safeguards carries not just reputational risk, but serious legal exposure — and that the people harmed by these failures are increasingly willing to fight back in court.
Conclusion
The class action lawsuits against xAI over Grok’s deepfake generation represent a critical inflection point in AI accountability. The core allegations — that the company knew its tool was being used to create millions of sexually explicit images, including thousands depicting minors, and chose to monetize the feature rather than disable it — strike at the fundamental question of what responsibilities AI companies owe to the public. The response from 35 state attorneys general and California’s cease-and-desist letter underscore that regulators view this not as a gray area but as a clear failure of corporate responsibility.
For victims of AI-generated deepfakes, these lawsuits offer a potential path to justice, but they also expose the limitations of the current legal framework. The cases against xAI will test untried legal theories, challenge existing platform liability protections, and potentially set precedents that shape AI regulation for years to come. Anyone who believes they may have been victimized by Grok-generated deepfakes should document all evidence, contact law enforcement, and consult with an attorney about their options — including whether they may qualify as members of a certified class if these suits move forward.
Frequently Asked Questions
Who can join the class action lawsuits against xAI over Grok deepfakes?
Both lawsuits seek class certification, which means they aim to represent all people similarly harmed by Grok’s deepfake generation. If the courts certify the class, individuals whose likenesses were used to create non-consensual deepfakes through Grok could potentially be included. Specific eligibility criteria will be defined by the court during the certification process.
Has xAI disabled Grok’s ability to generate explicit images?
Following California Attorney General Rob Bonta’s cease-and-desist letter on January 16, 2026, xAI restricted Grok’s image generation features — including “Spicy Mode” — to paying subscribers within 24 hours. However, critics and plaintiffs argue that restricting the feature to paid users rather than disabling it entirely does not constitute a meaningful fix.
What laws did California cite in its cease-and-desist letter to xAI?
California Attorney General Bonta cited violations of four California statutes in the January 16, 2026, cease-and-desist. The letter demanded that xAI halt the generation of illegal deepfake content, confirm compliance by January 20, 2026, at 5:00 PM PT, and preserve all related evidence.
How many deepfake images did Grok generate according to researchers?
The Center for Countering Digital Hate estimated that over a nine-day period, Grok generated more than 4.4 million total images, with 1.8 million being sexualized depictions of women. In an overlapping 11-day window from December 29, 2025, to January 8, 2026, researchers estimated over 3 million sexualized images were created, approximately 23,000 of which depicted minors.
Can the person who created the deepfakes also face criminal charges separate from the civil lawsuit?
Yes. The individual who allegedly created the deepfakes of the Tennessee teens was arrested in late December 2025 and faces separate criminal proceedings. Civil lawsuits against xAI and criminal charges against the individual perpetrator are independent legal actions that can proceed simultaneously.
What should I do if I discover that AI-generated deepfakes have been made using my likeness?
Document everything immediately — take screenshots of the images, note URLs and platform names, and record timestamps. File a report with local law enforcement and the FBI’s Internet Crime Complaint Center (IC3). Contact the platforms hosting the content to request removal. Consult with an attorney who handles privacy or digital exploitation cases, as you may have civil claims against both the person who created the images and the AI company whose tool was used.