Deepfake Abuse Class Action Against SpaceXAI Expands, Adding Stability AI

An amended complaint filed July 7 in the Northern District of California brings the number of Jane Doe plaintiffs to five and names the maker of Stable Diffusion as a defendant alongside Elon Musk's AI company.

Jul 10, 2026Source: NPR

Deepfake Abuse Class Action Against SpaceXAI Expands, Adding Stability AI

Key Points

1.Plaintiffs' firm Lieff Cabraser filed an amended complaint on Tuesday, July 7 in the U.S. District Court for the Northern District of California, adding plaintiffs from Wyoming and Wisconsin to the three Tennessee teenagers who sued in March, and naming Stability AI as a new defendant beside SpaceXAI, the company formerly known as xAI.
2.The complaint alleges that a stepfather used Grok to generate roughly 7,000 sexually explicit images and videos of 'Jane Doe 4' from a single photograph taken when she was about 11, and that SpaceXAI filed just one tip to the National Center for Missing & Exploited Children, in February, despite the scale of the material.
3.The suit brings claims under Masha's Law, the Trafficking Victims Protection Act and California law — including negligence, defective product design and public nuisance — and seeks injunctive relief plus damages for a proposed nationwide class.

The class action over AI-generated child sexual abuse material has widened. On Tuesday, July 7, lawyers at Lieff Cabraser Heimann & Bernstein filed an amended complaint in the U.S. District Court for the Northern District of California adding two plaintiffs and a second defendant. The case, first filed in March 2026 by three Tennessee teenagers, now has five Jane Doe plaintiffs and names Stability AI — the maker of Stable Diffusion — alongside SpaceXAI, the Elon Musk company known as xAI when the suit began.

The new allegations

The two plaintiffs added to the complaint are women from Wyoming and Wisconsin. According to the filing, the stepfather of the plaintiff identified as Jane Doe 4, now in her twenties, used Grok to generate roughly 7,000 sexually explicit images and videos of her, all derived from a single photograph taken when she was about 11 years old. The complaint says some of the material carried explicit captions.

The reporting claim

US companies are legally required to report suspected child sexual abuse material to the National Center for Missing & Exploited Children. The complaint alleges SpaceXAI submitted a single tip to NCMEC, in February, a number the plaintiffs argue is irreconcilable with the volume of imagery their clients say Grok produced.

The legal theory

The claims run under Masha's Law, the Trafficking Victims Protection Act and California state law, and include production of CSAM, benefiting from a sex-trafficking venture, negligence, defective product design and public nuisance. That last count is the widest. "Public nuisance means that this is not just something that is problematic for our clients," said Annika K. Martin, a Lieff Cabraser partner who heads the firm's Sexual Abuse Survivors Practice Group and leads the case with co-counsel Vanessa Baehr-Jones. "This is something that is a scourge on society." The proposed nationwide class covers anyone in the US whose real childhood images were altered by Grok into sexualized material. Plaintiffs seek damages and a court order forcing effective guardrails.

The responses

Stability AI said in a statement that "any suggestion that safety is not a top priority for us is categorically wrong," adding, "we take our ethical responsibilities seriously." SpaceXAI did not respond to requests for comment. Regulators reached the same subject first: Britain's Ofcom opened an Online Safety Act investigation into X on January 12, the ICO opened its own on February 3 over Grok's sexualized image generation, and the European Commission has opened proceedings under the Digital Services Act.

Why It Matters

By adding Stability AI, the plaintiffs are testing whether liability for synthetic abuse imagery extends past the consumer product to the open-weight model underneath it — the question that decides how much of the generative stack sits inside US courts' reach. The public-nuisance count and the NCMEC-reporting allegation aim at the same target: a legal duty to build guardrails before harm, not to answer subpoenas after it.

Filed underUS AI Regulation AI Safety #Grok #SpaceXAI #Stability AI #deepfakes