Over the last several months, the boards of a number of tech companies have been hit with “follow on” shareholder derivative lawsuits, after the companies were first sued in underlying intellectual property suits. The derivative lawsuits allege that the companies’ boards knowingly allowed their companies to use copyrighted materials to train their AI models, resulting in the underlying IP liability litigation, as well as potential IP-related liability. In the following guest post, Nathaniel French and Mason Dressler take a detailed look at the latest of these lawsuits, filed against Apple’s board. Nate is a partner and Mason is an associate at the Kennedys law firm. Our thanks to Nate and Mason for allowing us to publish their article on this site.
**************************
Intro
Did Apple’s directors and officers know that their teams developing Apple’s AI platform, Apple Intelligence, were utilizing copyrighted, unlicensed datasets of pirated books, movies, songs, voices, and YouTube videos? This is one of the key questions posed by Phil Rosen in his lawsuit filed against Apple.
This lawsuit marks a continuation in the trend of shareholders raising derivative lawsuits in which copyright infringement related issues are transformed into claims alleging that directors and officers breached their fiduciary duties. In the past three months, similar lawsuits have been filed against Microsoft’s board, Nvidia’s board, and Adobe’s board. It appears that these four lawsuits all follow a similar playbook in which the Plaintiff alleges that various pirated datasets were utilized in training AI technologies, that the board knowingly and intentionally permitted usage of these pirated datasets, and that the board should have taken more care in managing what data was utilized in training AI technologies.
Moreover, all of these lawsuits stem from the usage of the Books3 dataset, which will likely continue to appear in forthcoming, similar lawsuits. The Books3 dataset contained over 191,000 fiction and nonfiction books; it was originally produced to be used by non-commercial, open-source projects by independent developers, but as this tidal wave of litigation continues, it is clear that more and more companies may have utilized this data set in their development of AI tools.
While artificial intelligence may promise to reshape the future, thisnewly filed shareholder derivative action against Apple Inc. asks whether that innovation came at the expense of the company’s legal and ethical obligations. This lawsuit provides an opportunity for continued analysis of the exposure and risks being shaped by the ongoing development of artificial intelligence, especially as it pertains to directors and officers.
The D&O Lawsuit
Rosen alleges that in June of 2024, Apple began the development of its AI platform, Apple Intelligence. In order to adequately develop this AI platform, Apple needed to train its language models with vast amounts of data. Rosen contends that the datasets used to train these models contained copyrighted works. For example, Rosen alleges that Apple trained its AI with a dataset containing nearly 200,000 pirated works. Moreover, Rosen further alleges that Apple trained a video-capable version of its AI platform on more than 4 million YouTube videos obtained without authorization and developed commercial voice models by ingesting human speech recordings without obtaining releases from the speakers, allegedly violating Illinois’ Biometric Information Privacy Act. As previously discussed, this lawsuit marks a continuation of a recent trend in which directors and officers are being implicated for the utilization of copyrighted works used in the development of AI technologies.
Although Apple utilized many of the same datasets as those identified in previous derivative lawsuits, such as (1) the Books3 pirated dataset and (2) a dataset containing over 70 million video clips from YouTube which were allegedly extracted from YouTube without authorization, Rosen’s complaint further alleges that Apple utilized an additional dataset containing audio recordings obtained from thousands of hours of human speech recordings produced by various Apple applications. For example, Rosen alleges that Apple utilized information obtained from live voicemail, iMessage audio transcription, and Siri.
According to Rosen, this behavior comes amidst a backdrop of frequent and ongoing litigation against similar technology companies based upon the utilization of pirated materials during the development of AI technologies. Rosen cites to the $1.5 billion in damages paid by another AI development company, Anthropic, alongside similar BIPA lawsuits filed against Microsoft, Alphabet, and Meta Platforms, Inc., in bolstering his argument that Defendants should have known about the risks of using these datasets.
Many of these lawsuits stem from the groundbreaking $1.5 billion Anthropic settlement in December of 2025. This lawsuit broke the tidal wave of litigation related to the usage of pirated datasets in developing artificial intelligence models. As described by one of the authors involved in the litigation, the Anthropic settlement ultimately was “about copyright theft. And today the extent of this theft has been revealed. Thousands of jobbing writers around the world are affected. This is a landmark ruling and the tip of the iceberg. The Anthropic Settlement has made very clear that authors will not stand for the unlawful use of their work. The message to AI companies who steal authors’ works is clear: you need to pay.” As we have seen in the months after the Anthropic settlement, shareholders are growingly increasingly aware of the risks companies willingly incurred while pursuing AI innovation at light speed. These lawsuits also are following a similar gameplan in which a follow-on suit is filed shortly after the first copyright-based action.
Rosen contends that Apple’s executives and directors were aware of these issues and ignored what he characterizes as clear warning signs from similar litigation against other AI developers. The complaint alleges that Apple’s directors and officers owed fiduciary duties of care, loyalty, good faith, oversight, and legal compliance to Apple and its shareholders. According to the complaint, these duties required them to act in the company’s best interests, exercise prudent supervision over its operations, ensure compliance with applicable laws, maintain adequate internal controls, and provide accurate disclosures to investors. Rosen contends that the defendants breached these obligations by allegedly permitting the use of copyrighted works and biometric voice data in the training of Apple’s AI models, failing to prevent or remedy related compliance risks, and approving or allowing misleading statements concerning Apple’s AI practices, privacy protections, and risk management.
As a result of the Defendants’ behavior, Rosen contends that Apple is now required to “defend itself in copyright infringement and BIPA violation cases” and is “facing potentially massive liability and related costs” along with the potential loss of customers.
Discussion This lawsuit against Apple marks the continuation of significant challenges brought by shareholders due to the unlicensed utilization of pirated data sets and continues a recent trend of D&O related claims arising from follow-on litigation related to data used in training AI models. Rosen characterizes the behavior of Apple as constituting a model of “ask forgiveness not approval” wherein their rushed innovation without paying attention to ongoing risks associated with using pirated datasets. This allegation is nearly identical to allegations brought against Adobe’s board, signalling that potential Plaintiffs are beginning to utilize a standard playbook in which they transform copyright actions into derivative breach of duty actions brought against directors and officers. While this complaint is newly filed, it will be interesting to follow this lawsuit in tandem with the previous lawsuits filed against similar companies like Adobe and Microsoft.
Moreover, this lawsuit against Apple underlines an additional instance of follow-up litigation being brought after original copyright claims were filed. In September of 2025, copyrighted book authors filed their class action lawsuit against Apple, see Hendrix v. Apple, Inc. Then, in April of 2026, intellectual property video creators filed their class action lawsuit against Apple, see Ted Entertainment, Inc. v. Apple, Inc. Thus, this lawsuit is indicative of the fact that insurers can likely expect these derivative, follow-on lawsuits will be filed shortly after the original copyright claim.
The Apple complaint continues a trend of directors and officers facing potential liability in light of the materials used to train AI models, especially during a time in which technology companies are racing to develop better models faster, almost akin to the space race of the 1960s. It further highlights the ways in which shareholders are growing increasingly aware of the risks involved with development of artificial intelligence. As such, it is increasingly important for directors and officers to continually work to understand the scope and depth of artificial intelligence usage and development within their companies.