Adobe is facing a new class-action lawsuit that accuses the company of using a massive trove of pirated books to train its artificial intelligence models. The case, filed by the Oregon author Elizabeth Lyon, is the latest in a wave of similar copyright lawsuits targeting major tech companies.
Lyon claims that Adobe trained its AI, specifically its “SlimLM” models used for document assistance on mobile devices, on a dataset that includes nearly 200,000 pirated books. Her own books, she says, were among them.
Adobe has denied the allegations. The company says its AI was trained on a different, open-source dataset. But the lawsuit argues that this dataset is actually a “derivative” of another dataset known to contain the pirated books. In other words, Lyon is claiming that Adobe knew, or should have known, that it was using copyrighted material without permission.
This isn’t the first time this specific dataset of pirated books has been at the center of a legal battle. It has also been mentioned in lawsuits against Apple and Salesforce, showing that this is a widespread problem in the AI industry.
Lyon says she is “committed to vigorously prosecuting this action” and is seeking damages and a formal declaration that Adobe willfully infringed on her copyright. The case is another clear sign that the legal fight over the data used to train AI is far from over.
