Message from October 30th, this week Meta The company filed an application with the U.S. District Court to dismiss an allegation that it had illegally downloaded the seedsSexual ContentFor AI TrainingLitigation。
According to TorrentFreak, Strike 3 Holdings had previously discovered that Meta's company IP address contained illegal downloads of some of its adult films, while it claimed that Meta had covered up other downloads through a “secret network” of 2,500 “hidden IP addresses”. Strike 3 alleges that Meta stole pornographic material, secretly trained its undisclosed adult version of the AI model behind Movie Gen and seeks compensation for losses that may exceed $350 million (note: the current exchange rate is approximately RMB 2486 million)。
On Monday, Meta filed a motion to dismiss the motionThe allegations of Strike 3 are “based on speculation and implication”, stating that the company “has been described by some as `the copyright hooligans' and has specifically filed extortion suits”I don't know. Meta requested the court to dismiss all copyright-related claims, arguing that there was no evidence that the technocrat directed the downloading of approximately 2,400 adult films under the flag of Strike 3 and that there was no evidence that it was aware of the illegal activity。
Meta states that Strike 3 “has not provided any evidence that Meta used adult images or video training AI models, let alone intentionally”。
Meta spokesman said to the technology media Ars, "These allegations are completely false."
Meta argues that downloading is “personal use”
It is worth noting that the marked downloading behaviour spans seven years, starting in 2018. Meta argues that the AI project related to its “multi-modular model and generated video research” was initiated about four years later, and that it is therefore unreasonable to state that these downloads were used for AI training. Meta also points out that a more “manifest” loophole is that its provisions prohibit the creation of adult content, “which contradicts the premise that such material may be useful for Meta AI training”。
According to Meta, the available evidence “specifyly indicates that”Marked adult content downloaded by seed is for "private personal use"— Because of the very small number of downloads associated with Meta IP addresses and employees, “tens of works are downloaded intermittently each year and only one document is obtained at a time”。
In its submission, Meta states that “a more reasonable inference from this fragmented and uncoordinated behaviour is that different individuals download adult videos for personal use”
FOR EXAMPLE, UNLIKE THE BOOK AUTHORS WHO SUE AI FOR TRAINING IN THE USE OF THEIR WORKS TO MAKE UP A VAST DATA SET, IT IS NOT POSSIBLE TO FIND A WAY TO USE THEMThe behavior on Meta IP addresses is only downloaded about 22 times a yearI don't know. Meta argues that this is far from “the large-scale coordinated data collection that the plaintiff claims is required for effective AI training”。
also,Meta claims that these so-called acts cannot even reliably be associated with any Meta employee。
Meta wrote in the document: “Strike 3 does not identify any of the individuals who allegedly used these Meta IP addresses, nor does it state that they were employed by Meta or were involved in Meta's AI training, much less specify what was allegedly downloaded for the training of specific Meta models.”
At the same time, Meta states that “tens of thousands of employees, as well as numerous contractors, visitors and third parties, use Meta's network to access the Internet every day”. Thus, while the content of Strike 3 has been downloaded over the past seven years by “possibly one or more Meta employees”, the “same possibility” is the responsibility of “visitors, netters, contractors, suppliers, maintenance personnel or any combination of the above”。
Other so-called related acts include a Meta contractor who was accused of downloading adult content at his father's home, but Meta argues that these downloads “is also clearly an individual consumption”. Meta stated that the contractor was an “automated engineer” and that there was no basis in its position for obtaining AI training data. Meta claims that “there is no fact that reasonably links Meta to these downloads”
Meta wrote in the document: “The fact that these alleged seed downloads stopped at the time of the termination of its contract with Meta does not indicate that Meta knew or directed them.”
Meta criticizes the AI training term "ludicrous."
Of the Strike 3 indictments, Meta may be the most perplexed by allegations of “hidden IP secret networks”. Meta states that this constitutes Strike 3 “an additional problem that cannot be solved” and writes:If Meta wants to "hidden" certain alleged plaintiffs and third-party content downloads, why use the easy-to-trace MEta IP address to make hundreds of other downloads”
Meta argues, "The obvious answer is that Meta will not do this.""The whole AI training theory" of Strike 3 is "pure and unfounded."。
Finally, Meta notes that Strike 3 cannot claim that Meta should have better “regulated” illegal activities in its network. Meta argued that “it would be an extremely complex and intrusive task to monitor every document downloaded by all those using Meta's global network” and cited jurisprudence that Meta was required by law only to take “simple measures” to monitor such activities。
Meta hopes that the Court will find that Strike 3 has failed to establish any link between Meta and these illegal downloads. Strike 3 needs to respond within two weeks。
For Meta, winning this suit is not only about avoiding compensation, but also about defending its commitment not to generate increasingly regulated pornographic content through AI video tools. In its submission, Meta indicated that Strike 3 failed to provide evidence of Meta's use of content training AI “because there is no such evidence at all”。
The spokesperson for Meta said to Ars: “We do not want to see such content and have taken precautions to avoid training on such material.”