His Finance Textbooks Turned Up in a Pirate Library. Now He's Arguing AI Will Erase the Market for Teaching Materials.

Alan Marcus writes the books MBA students lug around. He is co-author of Investments, Essentials of Investments and Fundamentals of Corporate Finance, and he was the first recipient of the Mario J. Gabelli Endowed Professorship at Boston College's Carroll School of Management.

On Aug. 14 he filed a proposed class action in the U.S. District Court for the Southern District of New York alongside five current and former professors from across the country. The suit seeks to represent every textbook author whose copyrighted work OpenAI or Microsoft allegedly copied.

The piracy allegation is the part that gets the headlines. The plaintiffs allege OpenAI obtained their works from pirated shadow libraries such as LibGen, ran a script to strip copyright management information from them, and used them to train its GPT models. In the complaint's words, OpenAI reproduced the works by "downloading them from shadow libraries, copying them into datasets and computer memory" and converting them into training-ready formats.

The complaint says the authors discovered their materials had been used without authorization after The Atlantic published an investigation in August 2023.

The argument that should interest anyone selling knowledge

Skip the piracy claim for a second. The interesting part is the economics.

The complaint places particular emphasis on market harm, arguing that the textbook market differs from the market for trade books because textbooks are typically selected by instructors or institutions rather than by the students who use them — and that AI-generated educational content could be substituted for textbooks, diluting demand and undermining the market for producing updated educational materials.

That is a different claim than "you stole my book." It's "you built a machine that replaces the reason anyone would ever buy my book again."

The filing argues that generative AI threatens to "displace human-authored textbooks with cheaper or no-cost AI-generated textbooks or textbook equivalents". Schools and students are always looking to cut education costs, the authors say, and if institutions decide "good enough" AI-generated alternatives will do, nobody funds the work of producing educational materials in the first place.

Swap "textbook" for "course," "cohort program" or "certification," and you have the anxiety that runs through every creator Slack group on earth. The difference is that these six have written it into a federal complaint with a damages demand attached.

The licensing road not taken

The plaintiffs are not arguing that AI training is inherently illegitimate. They're arguing they were never offered a deal.

The suit says a licensing market for textbook content exists or could be developed that would allow textbooks to be used in training in exchange for "fair compensation," but that the defendants bypassed those licensing efforts, including by downloading massive amounts of textbooks from pirate sites — and that because textbooks cover academic fields in engaging, effective and up-to-date ways, they are "uniquely valuable" for training LLMs.

Microsoft is in the case as the infrastructure. The plaintiffs name it for building the infrastructure on which OpenAI trained its models and for profiting by building those models into products like Bing Chat and Copilot.

OpenAI has pointed to fair use. A spokesperson told amNewYork that "our models are trained on publicly available data and grounded in fair use".

Where it actually stands

The case is no longer a standalone fight. The Sullivan class action, led by mathematics author Michael Sullivan, has been consolidated into In re OpenAI, Inc. Copyright Infringement Litigation, MDL No. 3143, before Judge Sidney Stein. The consolidation lands as the U.S. government weighs in on whether training LLMs on copyrighted works is fair use.

The authors want a judgment of infringement, monetary damages and an injunction blocking "further infringement." The same group filed a similar suit against Meta in California in July.

If Stein takes the market-dilution theory seriously, it becomes the first serious legal test of whether "an AI can approximate my paid educational product" counts as a compensable harm. Every person selling a $997 course has a stake in that answer, whether or not they ever write a textbook.