Meta Seeks Dismissal of Strike 3’s $350 Million Copyright Lawsuit
Meta contests allegations that it downloaded thousands of adult films for AI-training, calling the claims speculative and unsupported
The technology giant Meta Platforms Inc. has formally asked a U.S. federal court to dismiss a lawsuit brought by adult-film producer Strike 3 Holdings, which alleges that Meta downloaded and distributed thousands of the company’s films via BitTorrent from 2018 onward in order to train its generative artificial intelligence models.
Strike 3’s complaint claims that approximately 2,400 of its titles were downloaded through Meta-owned corporate IP addresses, with an additional network of about 2,500 masked IPs alleged to have been used for the activity.
According to the complaint, Meta’s motive was to seed the films on peer-to-peer networks in order to leverage the “tit-for-tat” nature of BitTorrent to accelerate downloads of other content used for training AI models such as Meta’s Movie Gen. The complaint further alleges that Meta continued to share the films for days, weeks or months after acquisition, potentially resulting in statutory damages exceeding 350 million dollars.
In response, Meta argues that the claims are built on “guesswork and innuendo” rather than concrete proof of wrongdoing.
A Meta spokesperson stated that there is no evidence the company intentionally acquired or used the works for AI-training, that adult-content is expressly forbidden under its terms of service, and that the volumes alleged—on average about 22 downloads per year using Meta’s IP addresses—are inconsistent with a coordinated effort to build a massive training database.
Meta adds that the downloads could more plausibly be attributed to private use by employees, contractors or visitors, rather than a company-directed strategy.
The lawsuit highlights key chronological and factual challenges in the plaintiffs’ case: the alleged downloads began in 2018, several years before Meta publicly disclosed major efforts in generative-video AI; and the complaint does not identify any specific model that was trained using the files.
Meta emphasises that the company neither ordered nor was aware of the downloads in question, and that the burden remains on the plaintiff to link the activity directly to Meta’s corporate operations.
The case has been assigned to the U.S. District Court for the Northern District of California and is currently in the early case-management stage.
The dispute unfolds amid intense legal scrutiny of how major technology companies collect, use and distribute copyrighted content in the age of generative AI. While earlier suits by authors and visual artists looked at large-scale text and image scraping, this case is distinct in focusing on adult-film content and peer-to-peer distribution networks.
Meta’s motion to dismiss gives the company an opportunity to avoid costly discovery and potential damages.
If the court allows the suit to proceed, it could require Meta to open internal records regarding IP-address usage, employee download behaviour and AI-training datasets—a disclosure Meta seeks to avoid at this stage.
Strike 3, for its part, insists the matter is one of systemic disregard for copyright in AI development.
A decision on the dismissal motion is expected within the coming months, and the outcome may influence how courts treat copyright claims tied to AI-training in the future.
Strike 3’s complaint claims that approximately 2,400 of its titles were downloaded through Meta-owned corporate IP addresses, with an additional network of about 2,500 masked IPs alleged to have been used for the activity.
According to the complaint, Meta’s motive was to seed the films on peer-to-peer networks in order to leverage the “tit-for-tat” nature of BitTorrent to accelerate downloads of other content used for training AI models such as Meta’s Movie Gen. The complaint further alleges that Meta continued to share the films for days, weeks or months after acquisition, potentially resulting in statutory damages exceeding 350 million dollars.
In response, Meta argues that the claims are built on “guesswork and innuendo” rather than concrete proof of wrongdoing.
A Meta spokesperson stated that there is no evidence the company intentionally acquired or used the works for AI-training, that adult-content is expressly forbidden under its terms of service, and that the volumes alleged—on average about 22 downloads per year using Meta’s IP addresses—are inconsistent with a coordinated effort to build a massive training database.
Meta adds that the downloads could more plausibly be attributed to private use by employees, contractors or visitors, rather than a company-directed strategy.
The lawsuit highlights key chronological and factual challenges in the plaintiffs’ case: the alleged downloads began in 2018, several years before Meta publicly disclosed major efforts in generative-video AI; and the complaint does not identify any specific model that was trained using the files.
Meta emphasises that the company neither ordered nor was aware of the downloads in question, and that the burden remains on the plaintiff to link the activity directly to Meta’s corporate operations.
The case has been assigned to the U.S. District Court for the Northern District of California and is currently in the early case-management stage.
The dispute unfolds amid intense legal scrutiny of how major technology companies collect, use and distribute copyrighted content in the age of generative AI. While earlier suits by authors and visual artists looked at large-scale text and image scraping, this case is distinct in focusing on adult-film content and peer-to-peer distribution networks.
Meta’s motion to dismiss gives the company an opportunity to avoid costly discovery and potential damages.
If the court allows the suit to proceed, it could require Meta to open internal records regarding IP-address usage, employee download behaviour and AI-training datasets—a disclosure Meta seeks to avoid at this stage.
Strike 3, for its part, insists the matter is one of systemic disregard for copyright in AI development.
A decision on the dismissal motion is expected within the coming months, and the outcome may influence how courts treat copyright claims tied to AI-training in the future.
