Two more survivors and a second AI company have been added to a class action accusing a chatbot maker of enabling the creation of child sexual abuse material. Days later, that same company turned around and sued one of its own users.
Reviewed by Survivor Justice Alliance · Updated 2026-07-18
Figures drawn from the amended federal class action complaint and related court filings reported in early-to-mid July 2026.
A class action first filed in March 2026 against an AI chatbot maker over abuse imagery generated by its image tool has grown considerably. An amended complaint filed the week of July 7, 2026 in federal court adds two new anonymous plaintiffs, now identified in filings only as Jane Doe 4 and Jane Doe 5, bringing the total number of plaintiffs in the case to five. The amendment also names a second company as a defendant for the first time, an AI developer whose openly released image models are accused of having been trained on abuse material and then released to the public without adequate safeguards.
The core claim running through the case has not changed: that both companies built and distributed tools capable of generating sexually explicit images of real, identifiable minors from ordinary photographs, and that neither did enough to prevent that outcome or to respond adequately once it was reported.
According to the amended complaint, Jane Doe 4's stepfather used the chatbot's image generator to produce more than 7,000 abuse-related images, all built from a single genuine photograph taken when she was eleven years old. The filing alleges the platform was chosen specifically because it imposed fewer restrictions on this kind of image generation than competing tools.
Jane Doe 5's account, also newly added to the case, describes a different perpetrator using a photo taken at her eighth-grade graduation to generate similar abuse material. Neither new plaintiff's account has been tested at trial, and the companies named have not yet filed a formal response to the amended complaint, but both accounts are now part of the sworn record the court will eventually weigh.
Companies that operate platforms capable of generating sexual imagery are required by federal law to report suspected abuse material to the national clearinghouse that coordinates with law enforcement. The amended complaint alleges that while the chatbot maker did submit a report tied to Jane Doe 4's case in February, it provided only the original authentic photo as supporting evidence and withheld the thousands of generated images along with account and device information investigators say they needed to identify the person responsible.
The complaint further alleges that this was not an isolated lapse. It cites findings that the large majority of the company's reports to the clearinghouse in early 2026 were not actionable by investigators because the underlying identifying data had been omitted, a gap that plaintiffs argue effectively shielded users who generated abuse material from being tracked down.
Rather than only defending itself against the survivors' claims, the chatbot maker filed its own separate lawsuit on July 15, 2026 in a Texas federal court against a man it accuses of using the platform's image generator to create abuse material. The company's complaint describes the conduct as a deliberate scheme that exposed real people to harm while creating legal and reputational risk for the platform itself, and it seeks monetary damages, reimbursement of the legal costs the company has incurred defending against related survivor lawsuits, and a permanent order barring the man from ever using the service again.
The move is notable strategically. By pursuing its own user in court, the company can point to an active enforcement case as evidence that it takes misuse seriously, even as it continues to face the survivors' allegations that its own design choices and reporting practices made that misuse easier to carry out in the first place, and harder for investigators to trace back to a specific person.
The amended complaint's allegations against the second company center on its decision to release image-generation models as open, downloadable software rather than a tool the company directly controls. Plaintiffs argue that once such a model is released publicly, anyone can strip out safety restrictions and build a new application on top of it, including so-called nudify apps that exist specifically to generate sexualized images of real people without their consent.
That distinction, between a company that operates a closed platform it can monitor and one that releases open model weights it cannot fully control after release, is likely to become a central legal question as the case moves forward, with implications well beyond these two companies for how AI developers are expected to build in safeguards before, not after, a model reaches the public.
The Alliance does not represent any party in this litigation and does not provide legal advice. Anyone whose likeness has been used to generate abuse imagery, or who is unsure how to report AI-generated abuse material, should contact law enforcement and consult an attorney experienced in this emerging area of law.
The July amendment reshaped the case in several concrete ways. Here's what's new:
The Survivor Justice Alliance is an attorney alliance and advocacy organization, not a law firm; nothing here is legal advice. Attorney advertising. Referrals and consultations are free, and alliance attorneys work on contingency. Support is available 24/7 at the RAINN hotline, 800-656-4673.
It is a federal class action alleging that two AI companies built and released image-generation tools that were used to create sexually explicit images of real, identifiable minors, and that the companies failed to take adequate steps to prevent or properly report that misuse.
Reports submitted without key information, such as the actual generated images or data that could identify the account or device involved, generally cannot be used by investigators to locate the person responsible, even though a report was technically filed.
A direct enforcement action lets the company demonstrate it is actively responding to misuse, though it does not resolve the separate legal question of whether the company's own design and reporting practices contributed to the harm.
That is one of the central disputed questions in this case. Plaintiffs argue the company remains responsible for foreseeable misuse of a model it chose to release; the company is expected to argue it cannot control what third parties later build on top of publicly released code.