Meta Platforms has asked a US court to dismiss a lawsuit accusing it of illegally downloading and distributing thousands of adult content to train its artificial intelligence systems. The motion to dismiss, filed in the U.S. District Court for the Northern District of California, alleges there is no evidence to support these claims.
Meta claims in its filing that there is no evidence that its models contain or were trained using copyrighted material. The company also claimed the allegations were “nonsense and unsubstantiated.” The motion was first reported by Ars Technica, and Meta directly denied the allegations, calling them false.
The plaintiffs “have gone to great lengths to piece together this story through speculation and innuendo, but their claims are unpersuasive and unsupported by well-argued facts,” the complaint states.
Meta asks court to dismiss adult content copyright infringement lawsuit
The first complaint was filed in July by Strike 3 Holdings. The platform claimed that Meta has used the company’s hidden IP addresses to torrent approximately 2,400 adult movies since 2018, and claimed the move was part of the company’s broader efforts to build a multimodal artificial intelligence system. Strike 3 is an adult film hub in Miami that distributes content under brands such as Vixen, Tushy, and Blacked.
Mehta argues in his motion that the scale and pattern of the alleged downloads are inconsistent with Strike 3’s AI training theory. Over the past seven years, only 157 pieces of adult content produced by Strike were allegedly downloaded using Meta’s corporate IP addresses. This results in an average of approximately 22 downloads per year across 47 different IP addresses.
Mehta’s lawyer, Angela Dunning, characterized the case as a “poor and disorganized operation” by “disparate individuals.”
Dunning noted that the download was by an individual and not part of an effort by the tech giant to collect data for AI training, as Strike 3 claims. The motion also pushes back on Strike 3’s claims that Meta used more than 2,500 hidden third-party IP addresses, noting that Strike 3 did not verify the addresses and instead created loose correlations. One of the IP ranges was said to be associated with a non-profit organization in Hawaii and had no connection to Meta, and the other IP ranges had no identified owners.
Meta also claimed there was no evidence it knew about the alleged downloads or could have prevented them. He added that the company does not derive any profit from it and that it is not easy or required by law to monitor every file on its global network.
Dermot McGrath, co-founder of venture capital firm Ryzen Labs, said: “If Meta admits that its data was used in its models, it will have to claim fair use, justify the inclusion of pirated content, and expose itself to the discovery of its internal training and auditing systems.” He added that instead of denying how the data was used, Meta chose to deny that it had used the data at all. But Mr McGrath argued that if the court allowed the defense, it could open up a “big loophole”.
He argues that this “could effectively undermine the copyright protection of AI training data lawsuits” and that future lawsuits would require “stronger evidence of a company’s direction” and that companies would be better at hiding it.
Still, there are good reasons to use explicit content, and some companies are using it to develop safety and moderation tools. “Most major AI companies have ‘red teams’ whose job is to use harmful prompts to probe weaknesses in their models and try to force the AI to generate explicit, dangerous or prohibited content,” McGrath said.
Sharpen your strategy with mentorship and daily ideas – 30 days free access to our trading program