Meta, the parent company of Facebook and Instagram, has strongly denied allegations that it used pornographic material to train its artificial intelligence (AI) systems. The company claimed that any adult content downloads liked to its network were made by individual employees for personal reasons, not for AI research or development.
The controversy began when Strike 3 Holdings, a well-known adult film production company, filed a copyright infringement lawsuit against Meta. The company accused Meta of illegally using its copyrighted adult videos to train an AI video generator called Movie Gen.
Meta’s Legal Response
In a recent court filing, Meta requested that the US District Court dismiss the lawsuit, arguing that Strike 3’s claims are based on “assumptions and speculation.” Meta pointed out that Strike 3 has a history of being labeled a “copyright troll” for filing what it described as “extortion-based lawsuits.”
According to Meta, there is no evidence that it used adult videos for AI training or even instructed employees to download such content. The company emphasized that the alleged downloads, which date back to 2018, occurred long before Meta began its AI research on “Multimodal Models and Generative Video” — a timeline that makes Strike 3’s claims “impossible.”
Meta also noted that the supposed downloads amounted to around 22 videos per year, which the company said is too small to suggest organized AI data collection. “It is far more likely,” Meta stated, “that a few individuals downloaded adult videos for personal use.”
Ongoing Scrutiny Over Meta’s AI Practices
This lawsuit comes at a time when Meta’s AI systems are already under scrutiny. A recent Reuters investigation revealed that internal Meta rules once allowed AI chatbots to engage in inappropriate conversations, spread false health information, and even assist users in making racist comments. In response, Meta updated its AI policies to address these concerns.
In its legal defense, Meta reiterated that its terms of service strictly prohibit generating adult content using its AI tools. The company stated that such content “has no place” in AI model training and that Meta actively takes steps to avoid it.
Strike 3’s Allegations and Meta’s Next Move
Strike 3’s lawsuit also accuses Meta of using a “stealth network” of 2,500 hidden IP addresses to download over 2,400 adult films and is demanding $350 million in damages. Meta has denied all these claims, stressing that Strike 3 has failed to identify any individuals responsible or prove a connection between the downloads and Meta’s AI projects.
A Meta spokesperson told ArsTechnica, “We don’t want this type of content, and we take clear steps to ensure our AI models are not trained on adult material.”
Strike 3 Holdings now has two weeks to respond to Meta’s motion to dismiss the case. The outcome could have significant implications for future AI copyright disputes, especially as tech companies continue to face growing questions about how their AI models are trained and what kind of data is used.
By
