Meta Accused of Pirating Adult Content to Train AI Models

Meta pirated and seeded porn for years to train AI, lawsuit says

Porn sites may have blown up Meta's key defense in a copyright fight with book authors who earlier this year said that Meta torrented "at least 81.7 terabytes of data across multiple shadow libraries" to train its AI models. Meta has defeated most of the authors' claims and claimed there is no proof that Meta ever uploaded pirated data through seeding or leeching on the BitTorrent network used to download training data. But authors still have a chance to prove that Meta may have profited off its massive piracy, and a new lawsuit filed by adult sites last week appears to contain evidence that could help authors win their fight, TorrentFreak reported. The new lawsuit was filed last Friday in a US district court in California by Strike 3 Holdings -- which says it attracts "over 25 million monthly visitors" to sites that serve as "ethical sources" for adult videos that "are famous for redefining adult content with Hollywood style and quality." After authors revealed Meta's torrenting, Strike 3 Holdings checked its proprietary BitTorrent-tracking tools designed to detect infringement of its videos and alleged that the company found evidence that Meta has been torrenting and seeding its copyrighted content for years -- since at least 2018. Some of the IP addresses were clearly registered to Meta, while others appeared to be "hidden," and at least one was linked to a Meta employee, the filing said. Meta's AI training tactics may have harmed minors According to Strike 3 Holdings, Meta "willfully and intentionally" infringed "at least 2,396 movies" as part of a strategy to download terabytes of data as fast as possible by seeding popular high-quality porn. Supposedly, Meta continued seeding the content "sometimes for days, weeks, or even months" after downloading them, and these movies may also have been secretly used to train Meta's AI models, Strike 3 Holdings alleged. The porn site operator explained to the court that BitTorrent's protocol establishes a "tit-for-tat" mechanism that "rewards users who distribute the most desired content." It alleged that Meta took advantage of this system by "often" pirating adult videos that are "often within the most infringed files on BitTorrent websites" on "the very same day the motion pictures are released." These tactics allegedly gave Meta several advantages, making it harder for Strike 3 Holdings' sites to compete, including potentially distributing the videos to minors for free without age checks in states that now require them. "Meta specifically targeted Plaintiffs' content for distribution in order to accelerate its downloads of vast amounts of other content," the lawsuit said. And while Meta claimed that it "wrote a script to intentionally limit distributing popular books on BitTorrent," Strike 3 Holdings believes "discovery will likely show" Meta "continuously" distributed its adult videos specifically as a strategy to get around the BitTorrent protocol. So far, Strike 3 Holdings says it has documented at least five episodes in which Meta "hand-picked" adult videos from a specific site for "intense periods of distribution" to avoid seeding other content it was sourcing through BitTorrent. "The only reason to incur the server and bandwidth expense of remaining in a swarm for these long durations is to leverage the extended distribution as tit-for-tat currency in order to efficiently download millions of other files from BitTorrent," Strike 3 Holdings alleged. Strike 3 Holdings is seeking extensive damages and an injunction to permanently stop Meta from pirating its videos. The company also wants Meta to delete any stolen videos from its AI training data and existing AI models. The company alleged that Meta could use its high-quality copyrighted works -- which provide rare long cuts of "natural, human-centric imagery" showing "parts of the body not found in regular videos" and "unique" forms "of human interactions and facial expressions" -- to create a rival adult video generator that could "eventually create identical content for little to no cost." "Plaintiffs cannot compete against Meta when it ignores federal and state laws and offers Plaintiffs' works for free," Strike 3 Holdings alleged. "This will effectively eliminate Plaintiffs' future ability to compete in the marketplace" as well as its brands' "hard-earned reputations as respected and ethical sources for high-quality adult motion pictures by potentially allowing minors unfettered access to Plaintiffs' content against Plaintiffs' consent." Ars could not immediately reach Meta, Strike 3 Holdings' in-house lawyer, or book authors' lawyers for comment. Evidence may prove Meta seeded more content Seeking evidence to back its own copyright infringement claims, Strike 3 Holdings searched "its archive of recorded infringement captured by its VXN Scan and Cross Reference tools" and found 47 "IP addresses identified as owned by Facebook infringing its copyright protected Works." The data allegedly demonstrates a "continued unauthorized distribution" over "several years." And Meta allegedly did not stop its seeding after Strike 3 Holdings confronted the tech giant with this evidence -- despite the IP data supposedly being verified through an industry-leading provider called Maxmind. Meta also allegedly attempted to "conceal its BitTorrent activities" through "six Virtual Private Clouds" that formed a "stealth network" of "hidden IP addresses," the lawsuit alleged, which seemingly implicated a "major third-party data center provider" as a partner in Meta's piracy. An analysis of these IP addresses allegedly found "data patterns that matched infringement patterns seen on Meta's corporate IP Addresses" and included "evidence of other activity on the BitTorrent network including ebooks, movies, television shows, music, and software." The seemingly non-human patterns documented on both sets of IP addresses suggest the data was for AI training and not for personal use, Strike 3 Holdings alleged. Perhaps most shockingly, considering that a Meta employee joked "torrenting from a corporate laptop doesn't feel right," Strike 3 Holdings further alleged that it found "at least one residential IP address of a Meta employee" infringing its copyrighted works. That suggests Meta may have directed an employee to torrent pirated data outside the office to obscure the data trail. The adult site operator did not identify the employee or the major data center discussed in its complaint, noting in a subsequent filing that it recognized the risks to Meta's business and its employees' privacy of sharing sensitive information. In total, the company alleged that evidence shows "well over 100,000 unauthorized distribution transactions" linked to Meta's corporate IPs. Strike 3 Holdings is hoping the evidence will lead a jury to find Meta liable for direct copyright infringement or charge Meta with secondary and vicarious copyright infringement if the jury finds that Meta successfully distanced itself by using the third-party data center or an employee's home IP address. "Meta has the right and ability to supervise and/or control its own corporate IP addresses, as well as the IP addresses hosted in off-infra data centers, and the acts of its employees and agents infringing Plaintiffs' Works through their residential IPs by using Meta's AI script to obtain content through BitTorrent," the complaint said.

Lawsuit says Meta pirated and distributed porn to train its AI

A new lawsuit filed by two companies in the adult film industry claims that Meta has torrented copyrighted adult film content for years in order to train its AI models. Strike 3 Holdings, a pornographic film holding company, along with Counterlife Media, filed the lawsuit in a US Court District in California on Friday, alleging that Meta "willfully and intentionally" infringed at least 2,396 copyrighted movies since 2018. The company is the most active copyright litigant in the U.S., according to TorrentFreak, and regularly targets BitTorrent pirates. According to the suit, Meta downloaded adult film content "for purposes of acquiring content to train its Meta Movie Gen, Large Language Model ('LLaMA'), as well as various other Meta AI Models that rely on video training content." Meta is accused of downloading copyrighted works without permission and with the purpose of distribution "in order to accelerate its downloads of vast amounts of other content." Popular torrenting networks don't just allow users to download media files. Torrenting is a form of peer-to-peer file sharing, in which users download pieces of files from other users. The suit alleges that not only did Meta download copyrighted adult works, but that it made these files available to others, and benefited from doing so. Mashable has reached out to Meta for comment. Strike 3 Holdings, which says its sites attract over 25 million monthly visitors, is now seeking extensive damages that could amount to $359 million. And, according to Ars Technica, the company is also asking for an injunction to permanently ban Meta from using its content again. Meta has previously been sued over its AI model training, with a major copyright and fair use case filed against the tech giant by authors including Richard Kadrey, Sarah Silverman, Ta-Nehisi Coates, and Junot Diaz. Recently, Meta won a ruling against these authors, who claimed the company used their copyrighted works without permission. As Mashable's Cecily Mauran reported, "At the heart of these fights is a battle of values: the rights and livelihoods of artists versus technological innovation at all costs."

Lawsuit Claims Meta Pirated Vast Numbers of Dirty Movies to "Train Its AI"

Mark Zuckerberg's Meta has been accused of pirating -- we are not kidding -- a whole adult bookstore's worth of dirty movies to train its artificial intelligence models. First flagged by the blog TorrentFreak, a copyright company called Strike 3 Holdings and an adult film studio called Counterlife Media have filed suit against Meta, alleging that the tech giant torrented nearly 2,400 copyrighted skin flicks. All in the name of AI research, of course. According to the suit, which was filed in California federal court, Strike 3 and Counterlife discovered by tracing dozens of IP and email addresses that Meta began downloading and seeding their content via Bittorrent way back in 2018. Because of the seeding, the lawsuit alleges, the company formerly known as Facebook also allegedly engaged in "methodical and persistent distribution of those works" to other parties -- including, potentially, to minors. "Defendant has continuously infringed Plaintiffs' Works for years," the suit claims, "often infringing the very same day the motion pictures are released." Citing Richard Kadrey et al v. Meta Platforms, an ongoing lawsuit brought in 2023 by authors whose work Meta has since admitted to pirating, the adult companies said they were alerted to the Zuckerberg-owned tech monolith's torrenting activities in January 2025 through coverage of that lawsuit. Using various infringement analysis and IP tracking tools, Strike 3 and Counterlife found that some 47 IP addresses associated with Meta -- including "at least one residential IP address of a Meta employee" -- had downloaded their copyrighted content. Stranger still, the way the data moved suggested "non-human patterns," and "the acquisition of this content [may have been] for AI training data." Though there's no exact stated reason why Meta, which long ago banned nudity on its platforms, would want to pirate all that smut, Strike 3 and Counterlife wagered a guess in the suit. "Plaintiffs' Works provide natural, human-centric imagery, which shows parts of the body not found in regular videos, and a unique form of human interactions and facial expression," the lawsuit reads, employing some choice legalistic innuendo to talk around the subject at hand. "Plaintiffs' motion pictures contain extended scenes without director cuts which enable AI models to experience continuity in a way that cannot be derived from most television shows or mainstream motion pictures." In short, Strike 3 and Counterlife seem to be claiming that Meta might have used their content to train its AI video generators, like Meta Movie Gen, to recreate human movement in ways that other stolen data simply can't quite nail, if you catch our drift. "Such models will eventually create identical content for little to no cost," the suit alleges. "This will effectively eliminate Plaintiffs' future ability to compete in the marketplace." Along with the deletion of any copyrighted and pirated content and an injunction to permanently bar Meta from torrenting its work again, Counterlife and Strike 3 are seeking damages of up to $150,000 per stolen video. With 2,396 pieces of content on the line, those damages could go as high as $359 million, TorrentFreak notes. While it seems strange to consider a world where Meta was using all that raunchy content to train AI, the fact remains that the company has admitted to pirating other content. With this compelling evidence and Zuckerberg's unquenchable thirst to stay abreast of the latest tech trends, it seems conceivable that Meta actually did what's being claimed -- even if it didn't necessarily plan to release the money shot, so to speak.

Meta allegedly torrented porn to train its AI

Strike 3 Holdings initiated a lawsuit against Meta in a U.S. district court in California on Friday, alleging Meta engaged in torrenting and seeding its copyrighted adult video content since at least 2018 to accelerate the download of other data for AI model training, according to an Ars Tecnica report. Earlier in the year, book authors accused Meta of torrenting 81.7 terabytes of data from "shadow libraries" to train its AI models. Meta subsequently defeated most of these claims, asserting a lack of proof that it uploaded pirated data through seeding or leeching on the BitTorrent network. Despite this, the authors retain an opportunity to demonstrate that Meta may have profited from extensive piracy. The recent lawsuit filed by Strike 3 Holdings introduces evidence that could potentially support the authors' ongoing litigation. Strike 3 Holdings, which operates adult video websites, states its sites attract over 25 million monthly visitors and are recognized as "ethical sources" for adult videos, distinguished by their "Hollywood style and quality." Following the disclosure of Meta's alleged torrenting activities by the authors, Strike 3 Holdings reviewed its proprietary BitTorrent-tracking tools, which are designed to identify infringement of its video content. This review, according to Strike 3 Holdings, uncovered evidence indicating that Meta has been involved in torrenting and seeding its copyrighted material for an extended period, dating back to at least 2018. The filing specifies that some of the identified IP addresses were clearly registered to Meta, while others appeared to be "hidden." Additionally, at least one IP address was linked to a Meta employee. Strike 3 Holdings alleges that Meta "willfully and intentionally" infringed upon "at least 2,396 movies." This action was part of a purported strategy to download terabytes of data rapidly by seeding popular, high-quality adult content. Meta allegedly continued seeding this content "sometimes for days, weeks, or even months" after the initial download. Strike 3 Holdings also suggested these movies might have been covertly utilized to train Meta's AI models. The adult content provider explained to the court that the BitTorrent protocol functions on a "tit-for-tat" mechanism, rewarding users who distribute highly sought-after content. Meta is accused of exploiting this system by "often" pirating adult videos, which are "often within the most infringed files on BitTorrent websites," frequently on the "very same day the motion pictures are released." These alleged tactics provided Meta with several advantages, simultaneously hindering Strike 3 Holdings' ability to compete. This includes the potential distribution of videos to minors without age verification, particularly in states that now mandate such checks. The lawsuit explicitly states, "Meta specifically targeted Plaintiffs' content for distribution in order to accelerate its downloads of vast amounts of other content." While Meta previously claimed to have "wrote a script to intentionally limit distributing popular books on BitTorrent," Strike 3 Holdings anticipates that "discovery will likely show" Meta "continuously" distributed its adult videos specifically as a strategy to circumvent the BitTorrent protocol's inherent reciprocation mechanism. Strike 3 Holdings asserts it has documented at least five distinct instances where Meta "hand-picked" adult videos from a specific site for "intense periods of distribution" as a means to avoid seeding other content it was acquiring via BitTorrent. Strike 3 Holdings further alleged that "The only reason to incur the server and bandwidth expense of remaining in a swarm for these long durations is to leverage the extended distribution as tit-for-tat currency in order to efficiently download millions of other files from BitTorrent." Strike 3 Holdings is seeking extensive damages and a permanent injunction to prevent Meta from continuing to pirate its videos. The company additionally demands that Meta delete any purportedly stolen videos from its AI training data and existing AI models. Strike 3 Holdings alleged that Meta could potentially employ its high-quality copyrighted works -- characterized by "natural, human-centric imagery" featuring "parts of the body not found in regular videos" and "unique" forms of "human interactions and facial expressions" -- to develop a rival adult video generator. Such a generator could "eventually create identical content for little to no cost." Strike 3 Holdings stated, "Plaintiffs cannot compete against Meta when it ignores federal and state laws and offers Plaintiffs' works for free." The company further alleged that this situation "will effectively eliminate Plaintiffs' future ability to compete in the marketplace" and undermine its brands' "hard-earned reputations as respected and ethical sources for high-quality adult motion pictures by potentially allowing minors unfettered access to Plaintiffs' content against Plaintiffs' consent." In response to inquiries regarding the lawsuit, a Meta spokesperson informed Ars, "We're reviewing the complaint, but don't believe Strike's claims are accurate." To substantiate its copyright infringement claims, Strike 3 Holdings conducted a search of its "archive of recorded infringement captured by its VXN Scan and Cross Reference tools." This search reportedly identified 47 "IP addresses identified as owned by Facebook infringing its copyright protected Works." The data allegedly indicates "continued unauthorized distribution" spanning "several years." Strike 3 Holdings claims that Meta did not cease its seeding activities even after being confronted with this evidence, despite the IP data reportedly being verified by Maxmind, an industry-leading provider. The lawsuit further alleges that Meta attempted to "conceal its BitTorrent activities" through the use of "six Virtual Private Clouds," which constituted a "stealth network" of "hidden IP addresses." This claim suggests the involvement of a "major third-party data center provider" in Meta's alleged piracy. An analysis of these IP addresses reportedly revealed "data patterns that matched infringement patterns seen on Meta's corporate IP Addresses" and included "evidence of other activity on the BitTorrent network including ebooks, movies, television shows, music, and software." The documented patterns on both sets of IP addresses are described as "non-human," suggesting the data was intended for AI training rather than personal use, according to Strike 3 Holdings. A Meta employee was previously noted to have stated, "torrenting from a corporate laptop doesn't feel right." Despite this, Strike 3 Holdings additionally alleged finding "at least one residential IP address of a Meta employee" engaged in infringing its copyrighted works. This suggests a possibility that Meta may have directed an employee to torrent pirated data using an off-site residential IP address to obscure the data trail. Strike 3 Holdings did not identify the specific employee or the major data center mentioned in its complaint. In a subsequent filing, the company acknowledged the potential risks to Meta's business operations and its employees' privacy if sensitive information were to be publicly shared. In total, Strike 3 Holdings alleged that the available evidence demonstrates "well over 100,000 unauthorized distribution transactions" linked to Meta's corporate IP addresses. Strike 3 Holdings aims for this evidence to lead a jury to find Meta liable for direct copyright infringement. Alternatively, if the jury determines that Meta successfully distanced itself from the alleged infringement by utilizing the third-party data center or an employee's home IP address, Strike 3 Holdings seeks to charge Meta with secondary and vicarious copyright infringement. The complaint asserts, "Meta has the right and ability to supervise and/or control its own corporate IP addresses, as well as the IP addresses hosted in off-infra data centers, and the acts of its employees and agents infringing Plaintiffs' Works through their residential IPs by using Meta's AI script to obtain content through BitTorrent."

Strike 3 Sues Meta for Pirating Adult Films to Train AI

MediaNama's Take: The lawsuit against Meta for allegedly pirating adult films to train AI models marks a significant point in the global copyright crisis around generative AI. Strike Three Holdings has provided forensic data in their lawsuit, including Internet Protocol (IP) logs, Packet Capture (PCAP) data, and evidence of prolonged BitTorrent seeding, suggesting not accidental scraping but intentional, systematic infringement. If proven, this isn't just copyright violation; it's commercial exploitation masked as research and development. Meta, like many AI companies, has leaned on the argument that training data qualifies as fair use, especially when outputs don't directly substitute the original. Courts, as seen in Kadrey v. Meta, have so far leaned cautiously in its favor. But Strike 3 reframes the debate not as abstract data use, but as platform-level piracy at commercial scale If their allegations are true, this isn't incidental ingestion. It's redistribution as a means to train faster and cheaper. This case underscores a regulatory vacuum; current copyright laws don't anticipate large-scale data laundering across torrent networks. Nor do they compel AI developers to disclose training data, leaving copyright holders to reverse-engineer infringement. In the absence of strong transparency rules or licensing norms, companies are incentivized to scrape first, settle later. India's ANI vs OpenAI case and the EU's new AI copyright compliance rules show that global momentum is shifting toward enforcement. If courts accept Strike 3's framing, that infringing distribution and AI training are interlinked, it could force the industry to overhaul not just how it trains models, but how it acquires and accounts for data in the first place. A new copyright lawsuit has been filed against Meta by two adult film producers, Strike 3 Holdings and Counterlife Media, in a California Federal Court on July 23, 2025. The lawsuit alleges, "Since 2018, Defendant (Meta) has infringed at least 2,396 movies owned by Plaintiffs." It further states that the infringement is intentional. The "intentional" tag by the adult film companies is based on the allegation that Meta downloaded these works from "pirate sources for purposes of acquiring content to train its Meta Movie Gen, Large Language Model ('LLaMA'), as well as various other Meta AI Models that rely on video training content." The litigation, filed in the Northern District of California Court, alleges rampant copyright infringement by Meta using the BitTorrent protocol. BitTorrent is a peer-to-peer file-sharing protocol that utilizes the Internet for communication, often used to pirate copyrighted media. According to the lawsuit, "Using the BitTorrent protocol, Defendant is committing rampant copyright infringement by both downloading Plaintiffs' Works as well as engaging in methodical and persistent distribution of those Works to others." BitTorrent works by breaking large files into small chunks and distributing them across many users, or "peers." Instead of downloading a file from a single source, users download chunks from multiple peers simultaneously. The system uses a "tit-for-tat" mechanism to encourage sharing: users who upload (or "seed") more data to others receive faster download speeds in return. This reciprocal exchange creates a kind of currency: the network prioritizes users who contribute more, giving them faster access to the chunks they need. This decentralized method speeds up file transfers. The companies allege that Meta also utilized this tit-for-tat system by uploading their copyrighted films on BitTorrent to facilitate their own downloads of other training content. "Defendant also distributes Plaintiffs' content on BitTorrent in order to capitalize on the reciprocal, 'tit-for-tat' mechanism embedded within the BitTorrent Protocol. Due to the popularity of Plaintiffs' content within BitTorrent, Defendant has specifically targeted Plaintiffs' Works for distribution as currency to support its downloading of a vast array of other content necessary to train its AI models," the lawsuit reads. According to the lawsuit, Meta distributed these copies for days, weeks, and months after downloading them from BitTorrent. Strike 3 discovered the alleged piracy after reviewing court documents from a separate case (Kadrey v. Meta), where Meta employees admitted to using BitTorrent to download pirated books. Armed with this information, Strike 3 says it identified 47 IP addresses registered to Meta that had actively downloaded and shared its content over a multi-year period. The plaintiffs verified these IPs using MaxMind geolocation services and matched them to extensive PCAP data collected through the VXN Scan and the Cross Reference Tool. The VXN Scan and the Cross Reference Tool were used to detect and document Meta's activity on BitTorrent. VXN Scan downloaded parts of the files directly from Meta-owned IPs and recorded over 100,000 unauthorized transactions. The Cross Reference Tool extracted metadata from torrent files and tracked the presence of Meta's IPs in BitTorrent swarms. Both tools generated PCAP evidence to show Meta's active role in distributing the content. Strike 3 also analyzed "off-infra" IP ranges, servers allegedly set up by Meta to mask its piracy. These IPs showed infringement patterns similar to Meta's known addresses, such as synchronized downloads in multiple languages and identical content in varying resolutions. The lawsuit links these addresses to six virtual private clouds that Meta employees admitted to using in a related case. One IP traced to a Meta employee's home also appeared in coordinated download patterns. The suit claims Meta continued to distribute the content long after downloading it, which the plaintiffs say proves the company used the films not just for training but also as redistribution tools. Even after Strike 3 alerted Meta's legal team and shared its evidence, the alleged infringement persisted. "Meta's willful infringement was for commercial use," the plaintiffs argue, adding that the company prioritized speed and volume of downloads over legality. They contend that Meta intentionally selected their videos because of their high resolution, unique subject matter, and uninterrupted scenes, attributes valuable for training generative AI models. Meta recently won a copyright lawsuit regarding the use of pirated work for AI training -- Kadrey v. Meta. In that case, a group of authors sued Meta for using pirated books, downloaded via shadow libraries, to train its LLaMA AI models. The lawsuit argued that Meta's AI illegally copied and relied on copyrighted text. In June 2025, a U.S. federal court dismissed most claims, ruling that the authors failed to prove Meta's use diluted or harmed the market for their works. The judge acknowledged dilution as a valid legal theory but said the plaintiffs lacked evidence that LLaMA outputs replaced or substituted the original books. This precedent makes Strike 3's lawsuit more consequential. If Strike 3 can show that Meta's unauthorized use and redistribution of adult films diluted the paid market by flooding torrent networks or helping AI replicate its style, it may succeed where Kadrey failed. Other AI companies face similar copyright suits. Authors in the US are suing OpenAI, alleging that it trained its models on copyrighted books. Visual artists have also sued Stability AI, Midjourney, and DeviantArt for scraping millions of images. Getty Images also sued Stability AI in the UK before withdrawing copyright infringement allegations. In India, Asian News International (ANI) sued OpenAI in 2024, claiming ChatGPT accessed and reused copyrighted news without permission, a case that could reshape India's AI copyright framework.