Business Law Journal

More than a dozen copyright lawsuits have been filed against OpenAI alleging similar use of copyright-protected works to train the large-language models (“LLMs”) used by OpenAI’s ChatGPT.  While comedian Sarah Silverman has joined at least two of these actions and claims her copyright class action in federal district court in California was the first copyright complaint filed against OpenAI based on use of copyrighted works to train LLMs, New York Times v. OpenAI & Microsoft is arguably the most spectated case in mass AI copyright infringement today. 

The New York-based federal district court overseeing the New York Times v. OpenAI & Microsoft copyright infringement and trademark dilution case recently denied Silverman’s attempt to intervene in the later-filed New York Times case, throwing another wrench in her so far rocky crusade against OpenAI and setting up potentially varied rulings in separate courts over OpenAI’s use of copyrighted works to train LLMs.

New York Times (“NYT”) filed suit against OpenAI and Microsoft seeking billions of dollars for copyright infringement and trademark dilution, generally alleging that OpenAI built the LLMs used by ChatGPT by copying published NYT articles and works of others with disproportionate deference to NYT works, the training of which was derived from cloud computing services of Microsoft, including Bing.  Due to this disproportionate deference, NYT contends that outputs generated by AI models powered by these LLMs could provide users with almost an entirety of an NYT article verbatim with simple prompting from the user.  Additionally, NYT alleged that hallucinations (i.e., bugs) experienced by AI models, which yields content that is false, fabricated, misrepresented, and/or “offensive” and falsely designates NYT as the content source, harm NYT’s reputation and goodwill and dilute its trademarks. 

OpenAI publicly responded to the allegations in NYT’s complaint, characterizing the suit as “without merit” and claiming that “memorizing” and “regurgitating” near verbatim copies or significant portions of published works is rare and that the company is making progress towards fixing this bug.  OpenAI added that the prompts used by NYT to yield the outputs that closely match published NYT articles used as exhibits were intentionally manipulated and already included large portions of the articles. 

NYT retorted by arguing that OpenAI’s response essentially conceded OpenAI’s disproportionate reliance on published NYT articles to train its AI models.  NYT argued that OpenAI’s use of NYT articles is not fair use, partly because it effectively replaces NYT and diverts NYT’s revenue sources to OpenAI. 

Silverman previously filed a similar copyright class action lawsuit against OpenAI in federal district court in California alleging, among other things, that OpenAI copied copyrighted works to train AI LLMs.  The court initially granted OpenAI’s motion to dismiss Silverman’s claims of vicarious infringement, Digital Millennium Copyright Act (“DMCA”) violation for removing copyright management information (“CMI”), and negligence.  The court reasoned that Silverman (a) failed to show direct copying (a requisite for alleging vicarious infringement) or substantial similarity between ChatGPT outputs and the copyrighted works; (b) showed no facts supporting removal of CMI by OpenAI; and (c) failed to show the existence of a duty of care by OpenAI to safeguard copyrighted works.  Still, the court preserved Silverman’s core claim of direct copyright infringement against OpenAI for training AI models based on her copyrighted works. 

Then, Silverman filed a motion to intervene the New York Times v. OpenAI & Microsoft case along with other similar cases, including the case brought by the Authors Guild, in an attempt to dismiss, stay, or transfer the case pursuant to Federal Rule of Civil Procedure 24 and the first-to-file rule.  Microsoft and NYT both opposed the motion while OpenAI took no position.  In denying Silverman’s motion to intervene, the court held that Silverman neither met the requirements of intervention as a matter of right under Rule 24(a) nor the requirements of permissive intervention under Rule 24(b). 

Rule 24(a) states that a party may intervene as of right on a “timely motion” showing that it “claims an interest relating to the property or transaction that is the subject of the action and is so situated that disposing of the action may as a practical matter impair or impede the movant’s ability to protect its interest, unless existing parties adequately represent that interest.”  Silverman’s claimed interest was to avoid potentially contradictory rulings between the NYT action and her case in California.  The New York federal district court found this insufficient, noting the differences between the two cases and the lack of class certification, which meant each Silverman plaintiff only had an interest in their specific actions, citing to instances of courts denying intervention due to the speculative nature of the interest before class certification.

The court also found that Silverman failed to demonstrate that the Silverman plaintiffs’ interests would be impaired or that NYT and other existing parties would not adequately represent their interests.  The shared goal of holding OpenAI accountable did not substantiate the need for intervention, especially given the absence of evidence of inadequate representation by the current parties.  Accordingly, the court denied Silverman intervention as a matter of right. 

Next, the court addressed permissive intervention under Rule 24(b), which allows for intervention if the party has a claim or defense sharing common legal or factual questions with the main action with the caveat that such intervention does not unduly delay or prejudice the original parties' rights.  The court’s decision paralleled the denial of intervention as a matter of right, highlighting a lack of cognizable interest, no impairment of interest, and adequate representation by existing parties, such as NYT.

Last, the court emphasized that allowing Silverman to intervene with the intent to dismiss, stay, or transfer the actions would prejudice the original parties, disrupting the established timeline and proceedings.  The court viewed Silverman’s motives for intervention unfavorably, which the court found focused on procedural adjustments rather than substantive contribution to the case.  The court prioritized the smooth progression of the New York Times case already underway.  The court’s decision is yet another early blow to Silverman’s strategy.  As a result, the Silverman’s action and the New York Times action will continue on parallel paths on both coasts, leaving different courts to each consider and independently decide overlapping novel legal issues, which may temper the several snowballing plaintiffs’ enthusiasm to bring new copyright infringement claims against OpenAI and Microsoft until certain keys issues are decided by the courts.

AALRR has a dedicated group of attorneys on its Intellectual Property Team with the experience and expertise to protect and vigorously enforce your copyrights and defend you against claims of copyright infringement.  Attorneys on the Firm’s Intellectual Property Team can also assist you with registration of your copyrights with the United States Copyright Office.  If you have questions about copyright protection or infringement of your copyrighted works, contact the authors or another member of the AALRR Intellectual Property Team. 

This AALRR post is intended for informational purposes only and should not be relied upon in reaching a conclusion in a particular area of law. Applicability of the legal principles discussed may differ substantially in individual situations. Receipt of this or any other AALRR information does not create an attorney-client relationship. The Firm is not responsible for inadvertent errors that may occur in the publishing process.   © 2024 Atkinson, Andelson, Loya, Ruud & Romo