Meta has firmly denied allegations that it used pirated pornographic videos to train its AI models, insisting that the accusations are baseless and unsupported by evidence. The controversy arose when Strike 3 Holdings, an adult film production company, filed a lawsuit claiming Meta downloaded nearly 2,400 copyrighted adult movies through hidden IP addresses to train AI systems, including a video generator called Movie Gen. However, Meta responded by stating that there is no proof that the company directed employees to download such content for AI training. Instead, the company argued that any downloads of adult videos on its network were sporadic, limited in number (around 22 per year), and likely made by diverse individuals for personal use rather than corporate AI development.

According to Meta, the downloads began as early as 2018, before the company started working seriously on AI models related to video generation. The company also emphasised that its internal policies prohibit training AI on adult content, contradicting the lawsuit’s premise. Meta’s motion to dismiss the suit highlights that Strike 3 failed to identify who made the downloads or whether the content was employed in any AI training, labelling the claims as speculative and defamatory. Meta further criticised the lawsuit as extortionate and relying on guesswork rather than concrete facts, calling for the court to dismiss the case.

The legal battle underscores growing scrutiny around how major tech firms acquire data for AI training and the potential copyright and ethical implications. While lawsuits against Meta and other AI firms like OpenAI continue over data sourcing, Meta maintains its innocence in both the specific case of adult content and broader AI training practices. The company insists on its commitment to responsible AI development and adherence to copyright and data privacy standards.

This case is especially sensitive for Meta, given its vast user base, including minors, and its ongoing efforts to incorporate AI technologies responsibly into its platforms. The lawsuit seeks over $350 million in damages, but Meta’s stance is clear: it neither intentionally nor knowingly used pornographic material in AI training, and the claims do not reflect its corporate conduct or policies. Meta awaits the court’s ruling on its motion to dismiss while continuing to develop and regulate its AI projects with transparency and caution.