Grok AI Deepfake Lawsuit: Cases, Investigations, and xAI’s Defense

The Grok AI deepfake lawsuits are a cluster of cases filed in early 2026 accusing Elon Musk’s xAI of building its Grok image tool without the safety guardrails other AI companies use, then profiting as users flooded X with sexually explicit fake images of real people, including children. At least four federal suits and one municipal enforcement action are now pending. None has reached trial. As of mid-2026, every case is still in its preliminary procedural stage, and the most active fight is over whether the victims can keep their names out of the public record.

What Grok Did

On December 29, 2025, xAI rolled out a feature letting X users edit any image on the platform with a single click. The Center for Countering Digital Hate later reported that Grok produced more than 4.6 million images in its first eleven days, roughly 3 million of them photorealistic sexualized images, generated at about 190 per minute. Researchers estimated some 23,000 depicted children.¹

Targets ranged from private individuals to Taylor Swift, Selena Gomez, Billie Eilish, and former Vice President Kamala Harris. On January 9, 2026, X restricted the image-editing feature to paid subscribers rather than disabling it. On January 14, Musk said he was “not aware of any naked underage images generated by Grok” and attributed the problem to “adversarial hacking.”² A day later, researchers still found 29 percent of the sexualized images of children they had identified publicly accessible on X.¹

Those facts form the common factual backbone of every lawsuit that followed.

St. Clair v. xAI

Ashley St. Clair filed the first suit on January 15, 2026, originally in New York state court. The complaint alleges that Grok generated sexually explicit deepfakes of her, including images based on a photograph taken when she was fourteen. She asserts negligence, emotional distress, and design defect claims, and alleges xAI retaliated against her by demonetizing her account, stripping her verification badge, and banning her from its premium subscription service.³

xAI responded by removing the case to the U.S. District Court for the Southern District of New York and filing its own lawsuit against St. Clair in federal court in Texas, claiming she violated its terms of service and seeking more than $75,000 in damages. The company’s position across every case is that claims against it must be filed in the Northern District of Texas or Tarrant County, Texas.⁴

Doe v. xAI Corp. — The Class Action

On January 23, 2026, a pseudonymous plaintiff identified as “Jane Doe” filed a proposed class action in the U.S. District Court for the Northern District of California. The case eventually grew to four pseudonymous plaintiffs — South Carolina Doe, South Carolina Roe, New Jersey Doe, and Ohio Doe — and now asserts eleven causes of action, including product liability, negligence, public nuisance, privacy violations, defamation, and unfair business practices.⁵

The complaint reaches directly into xAI’s design choices. It claims the company’s own system prompts told Grok that if content was not flagged within designated policy tags, the chatbot had “no restrictions on adult sexual content or offensive content.” Plaintiffs allege xAI “expressly programmed Grok to make any ‘adult sexual content’ requested by a user” and abandoned standard safety testing. They further allege that moving the tool behind a paywall after the scandal was an effort to “further exploit the women victimized by the trend for additional commercial profit.”⁵

The specific harms described include images of South Carolina Doe in a revealing bikini she would never have shared publicly, deepfakes of New Jersey Doe in sexually explicit poses created after he asked the chatbot not to generate images of him, and images of South Carolina Roe as a child digitally altered to depict her in bed with her father in a way implying sexual abuse.⁶

The Tennessee CSAM Suit

On March 16, 2026, three Tennessee teenagers filed a separate suit in the Northern District of California. One was a recent adult who was a minor at the time of the alleged conduct; two remain minors. The case concerns a third-party application that accessed xAI’s “Grok Imagine API” and used it to generate and distribute sexualized deepfakes of the plaintiffs, including nude depictions. The complaint alleges the images were traded on Discord and Telegram as currency for other child sexual abuse material.⁷

The plaintiffs assert thirteen counts, including claims under Masha’s Law and the Trafficking Victims Protection Act, as well as negligence, intentional infliction of emotional distress, public nuisance, and the production, possession, and distribution of child pornography. They seek compensatory and punitive damages and class-action status on behalf of anyone in the United States “reasonably identifiable” in sexualized content Grok created using their real likeness.⁸

A criminal investigation preceded the civil case. Police arrested a suspect in December 2025 after one victim alerted the others and their families, and investigators reportedly recovered AI-generated images on the suspect’s phone that had been morphed to depict the victims nude.⁹

The Anonymity Fight

The procedural battle most likely to shape the class action right now is over whether the plaintiffs can keep their names sealed. A federal judge initially approved pseudonymity for the lead plaintiff. After the case was refiled in early May 2026 with four pseudonymous plaintiffs, xAI filed two motions in mid-May seeking to overturn that ruling.⁶

xAI’s lawyers, from Cahill Gordon and Reindel and Sawyer and Labar, argue that federal procedural rules generally require parties to be named and that the public has an interest in knowing who is suing the company. In a May 15 filing, they wrote that “factoring out the deepfake image itself — as it will remain under seal — there is nothing inherently stigmatizing about revealing the fact that a deepfake image was created.”⁶

The plaintiffs, represented by Sophia Rios of Berger Montague, responded that unmasking would expose them to harassment, doxing, and retaliatory creation of more extreme deepfakes. Their opposition called the motion an “obvious effort to intimidate Plaintiffs into dropping the litigation.” All four submitted affidavits stating they would consider abandoning the case entirely if forced to reveal their names. A hearing is scheduled for July 9, 2026, before Judge P. Casey Pitts.⁶¹⁰

Baltimore’s Consumer Protection Suit

On March 24, 2026, the mayor and city council of Baltimore became the first municipal government to sue xAI over Grok. Filed in Baltimore City Circuit Court, the complaint names X Corp., x.AI Corp., x.AI LLC, and SpaceX as defendants and alleges violations of Baltimore’s Consumer Protection Ordinance.¹¹

Baltimore’s theory is different from the private plaintiffs’. Rather than seeking damages for specific victims, the city alleges deceptive trade practices: that xAI marketed X and Grok as safe, rule-governed products while failing to disclose the tools’ potential for generating nonconsensual intimate imagery and child sexual abuse material. The complaint points to the gap between X’s stated policies, which prohibit nonconsensual sexual content and child exploitation, and what Grok actually produced. Baltimore seeks the maximum statutory penalties per violation, injunctive relief, and restitution. The complaint also characterizes Musk’s own December 31, 2025 post of an edited bikini image of himself as a public endorsement of Grok’s “undressing” capabilities.¹²

Parallel Government Investigations

Two regulators are moving on xAI alongside the civil cases. On January 14, 2026, California Attorney General Rob Bonta opened an official investigation into xAI and followed it two days later with a cease-and-desist letter. Bonta described the creation of child sexual abuse material as a crime and said xAI’s practices violated California civil laws.¹³

On January 12, 2026, the UK media regulator Ofcom opened a formal investigation into X under the Online Safety Act, examining whether the platform failed to assess risks of illegal content, failed to prevent users from encountering child sexual abuse material and intimate image abuse, and failed to implement effective age-gating for pornographic content. If Ofcom finds a breach, it can impose fines of up to ten percent of qualifying worldwide revenue or seek court orders blocking UK users from the platform entirely.¹⁴¹⁵

How xAI Is Defending Itself

xAI’s defense so far has been procedural rather than substantive. The company has pushed to move cases to the Northern District of Texas under its terms of service, has moved to strip the Doe plaintiffs of anonymity, and has argued that because the deepfake images themselves will remain under seal, no compelling privacy interest justifies pseudonymity.⁶

Early reporting indicated xAI intended to seek dismissal under Section 230 of the Communications Decency Act, which shields platforms from liability for content posted by users. Whether that defense works is an open question. The legislative authors of Section 230 have said they do not believe its liability shield covers generative AI systems, and in oral arguments in an earlier Supreme Court case, Justice Neil Gorsuch suggested AI-generated content might fall outside the statute’s protections.¹⁶ The core question is whether xAI is hosting user content or generating it. As of June 2026, no Section 230 motion to dismiss has actually been filed in any of the cases.¹⁰

SpaceX, which now owns xAI, has reportedly set aside more than $500 million to address fallout from the Grok deepfake litigation. Neither xAI nor its legal team has commented publicly on the substance of the claims.⁶

Where the Cases Stand

As of mid-2026, none of the deepfake lawsuits against xAI have moved past preliminary procedure. No class has been certified. No motion to dismiss has been filed on the merits. No trial date has been set. The July 9, 2026 hearing on anonymity in the Doe class action is the next scheduled milestone. The California attorney general’s investigation and Ofcom’s UK inquiry both remain open.¹⁰

The underlying legal question running through all of the cases is whether a generative AI company is a content creator, potentially liable for what its model produces, or a platform entitled to Section 230 immunity. How courts answer that will reach well beyond Grok. For now, the plaintiffs are fighting to keep their names sealed while pressing claims that test the outer edges of product liability, privacy law, and child protection statutes.

1
Center for Countering Digital Hate. Grok Floods X With Sexualized Images
2
The 19th. Women, Girls Lawsuit Grok AI Deepfakes
3
Forbes. Ashley St. Clair Sues xAI Over Sexualized Deepfakes
4
NBC News. Ashley St. Clair Sues xAI Over Grok Sexual Images
5
Tech Policy Press. Breaking Down a Class Action Lawsuit Filed Over Grok Undressing Controversy
6
WIRED. xAI Asks Court to Strip Alleged Grok Deepfake Nudes Victims of Anonymity
7
The Guardian. Lawsuit Elon Musk AI Grok Child Sexual Abuse
8
The Commercial Appeal. Elon Musk Grok xAI Lawsuit Tennessee Child Sexual Deepfakes
9
Business Insider. Lawsuit Allegation xAI Grok Created Sexualized Deepfakes Minors
10
CourtListener. Doe v. xAI Corp., Docket
11
Courthouse News Service. Baltimore Sues X, Elon Musk AI Generated Porn
12
NBC News. Baltimore Sues Musk’s xAI Over Grok’s Sexually Explicit Images
13
Axios. xAI California Elon Musk Deepfakes Children Grok
14
The Guardian. Ofcom Investigating X Outcry Sexualised AI Images Grok Elon Musk
15
UK Government. Secretary of State Statement to the House of Commons
16
Stanford HAI. Does Section 230 Cover Generative AI