Meta Sued for Allegedly Using Pirated Pornographic Content for AI Training

• Meta copyright lawsuit: Two adult film producers, Strike 3 Holdings and Counterlife Media, accuse Meta of training AI on pirated pornographic videos.
• New accusations: The plaintiffs claim that the company illegally downloaded and shared nearly 2,400 films to train Meta Movie Gen and LLaMA.
• Pirated content training: This is not the first time Facebook’s parent company has faced legal battles for its AI training methodologies.

A new Meta copyright lawsuit has been filed by adult entertainment companies Strike 3 Holdings and Counterlife Media, which accuse Meta of direct and secondary copyright infringement. They claim that Meta utilized 2,396 pirated films to train its AI models, including Meta Movie Gen and LLaMA.  

Allegations and Evidence  

According to court documents submitted in California, Meta allegedly downloaded these films from BitTorrent networks, without authorization, with potential damages running up to $359 million.

The companies claim the content, which belongs to brands like Vixen and Tushy, was integral to training AI systems capable of generating video content resembling "Hollywood-grade" production quality. 

Strike 3 and Counterlife Media searched for Meta-linked IP addresses in their collected BitTorrent data archive and identified 47 reportedly infringing IPs, which were reportedly owned by Facebook.

Evidence provided includes IP address data linked to Meta’s infrastructure and third-party proxies, suggesting deliberate actions to engage in copyright infringement for accelerated downloads, the complaint says.

The complaint revealed that a Comcast IP address was used by an identified Facebook employee with a redacted name to download content, who allegedly shared content via Meta corporate IPs and covert IP addresses.

Implications and Legal Ramifications  

The copyright infringement case underscores the increasing scrutiny over how companies source data for AI development.  

Last week, Meta won a lawsuit in which 13 authors, including Richard Kadrey, Sarah Silverman, and Christopher Golden, claimed that Facebook’s parent company downloaded 81TB from shadow libraries for AI training. The judge limited the ruling to the authors’ case, saying the plaintiffs “made the wrong arguments.”

Critics argue that the use of copyrighted material for AI without proper licensing sets a dangerous precedent, potentially harming creators' intellectual property rights while also raising ethical concerns about transparency in AI training.