, ,

NVIDIA asserts Millette lacks standing due to failure to alleged “a concrete, particularized injury in fact” under Constitution, simply based on scraping YouTube videos to train AI

NVIDIA filed its motion to dismiss David Millette’s complaint alleging state law claims for NVIDIA’s alleged use of his videos to train NVIDIA’s AI model. NVIDIA leads off with the argument that Millette’s bare allegations do not satisfy the U.S. Constitution’s Article III requirement of a concrete, particularized injury to establish a plaintiff’s standing in federal court.

The Complaint lacks even the most basic information, such as the name of Plaintiff’s YouTube account; what videos he uploaded; and when between 2009 and 2024 they were uploaded. Plaintiff’s threadbare assertions and conclusions not only fail to allege any particularized harm to Plaintiff, but they also fail to “give the defendant fair notice of what the claim . . . is and the grounds upon which it rests.” Twombly, 550 U.S. at 555 (quotation marks omitted).

Further, Plaintiff provides no facts to support a plausible inference that his videos were allegedly scraped or used by NVIDIA. There are billions of videos currently on YouTube. Yet Plaintiff asserts only that NVIDIA has “downloaded 100,000 videos,” Compl. ¶ 20, “scrapped [sic] 16  millions of YouTube videos,” id. ¶ 3, and “compiled 38.5 million URLs.” Id. He thus provides no plausible basis to conclude that this tiny fraction of the content on YouTube includes his videos.

Recent precedent makes clear that the mere possibility of Plaintiff’s video content being scraped is insufficient to confer standing. For example, Brantley v. Prisma Labs, Inc., No. 23 C 1566, 2024 WL 3673727, at *2, 5 (N.D. Ill. Aug. 6, 2024), recently dismissed on standing grounds a claim that a plaintiff’s photographs were among billions allegedly scraped without consent from social media sites to train an AI model. Although the model at issue in Brantley purportedly scraped “almost every website from September 2021 to January 2022,” the plaintiff provided no additional facts to establish that his photos “were contained in the [relevant] [d]ataset.” Id. at *5 (quotation marks omitted). The complaint, therefore, pled “facts that [were] merely consistent with a defendant’s liability, [but that] stop[ped] short of the line between possibility and plausibility of entitlement to relief.” Id. at *6 (internal quotes omitted); Doe 1, 2024 WL 235217, at *5 (dismissing for lack of standing where certain plaintiffs “failed to plead specific instances in which their code” was unlawfully reproduced by defendant’s AI model); cf. In re Google Assistant Priv. Litig., 457 14  F. Supp. 3d 797, 816 (N.D. Cal. 2020) (dismissing Stored Communications Act claim where plaintiff failed to plausibly allege that, among 153 unlawfully intercepted recordings, “Plaintiffs’ own oral communications were intercepted”).

I think this standing issue will become a big issue in the AI copyright litigation. What’s interesting is that, thus far, only NVIDIA has raised Article III standing as a basis for dismissal in the 3 lawsuits brought by Millette. But Judge Davila presides over all 3 cases brought by Millette. Courts can raise the issue sua sponte if not raised by the parties. See Adarand Constructors, Inc. v. Mineta, 534 U.S. 103, 110 (2001) (“We are obliged to examine standing sua sponte where standing has erroneously been assumed below.”). So, we can expect Article III standing will be an issue in all 3 Millette lawsuits that Judge Davila will consider.

Millette v. NVIDIA: NVIDIA filed its motion a few days before Judge McMahon’s recent decision dismissing Raw Story Media’s complaint alleging a federal claim based on Section 1202 of the DMCA against OpenAI; Judge McMahon dismissed the complaint precisely for lack of standing due to the failure to identify a concrete injury suffered from mere alleged use of the plaintiffs’ works with copyright management information removed to train OpenAI’s model. As Judge McMahon concluded, “I am not convinced that the mere removal of identifying information from a copyrighted work—absent dissemination—has any historical or common-law analogue.”

Millette v. Google: Google didn’t raise this Article III standing argument in its motion to dismiss the similar complaint filed by Millette against it. Instead, Google raises a statutory standing challenge under the Unfair Competition Law of California.

Millette and Petryazhna v. OpenAI: Although Millette’s lawsuit against OpenAI started out alleging the same state law claims against OpenAI, Millette filed a first amended complaint adding a copyright claim based on the additional plaintiff Plaintiff Petryazhna and her videos. I assume at least one of the reasons for the additional plaintiff was her registrations of her videos in the Copyright Office, a prerequisite for a copyright infringement claim (see First Am. Complaint, para. 12). Based on the information in the docket, however, it’s unclear why the same strategy has not (yet) been pursued by Millette in the other two lawsuits, which are almost identical.

Leave a Reply


Discover more from Chat GPT Is Eating the World

Subscribe now to keep reading and get access to the full archive.

Continue reading