Adobe has become the latest technology heavyweight pulled into court over how artificial intelligence systems are built, facing a proposed class-action lawsuit that alleges the company trained an AI model using pirated books — including works written by the lead plaintiff.

The Adobe AI pirated book suit intensifies a growing legal clash over whether AI developers crossed legal boundaries while feeding massive datasets into their models. It also deepens scrutiny of how generative AI tools are trained and whether creators’ rights were sidelined in the race to innovate.

Author Claims Her Books Were Used Without Permission

The lawsuit was filed by author Elizabeth Lyon, who alleges Adobe used pirated copies of her nonfiction books to train its SlimLM language model. Adobe has described SlimLM as a compact model designed for document-related tasks, particularly on mobile devices.

According to the complaint, SlimLM was trained using a dataset called SlimPajama-627B, released by AI chipmaker Cerebras in June 2023. That dataset, the lawsuit says, is derived from an earlier collection known as RedPajama, which itself incorporates material from the Books3 dataset.

Books3 contains more than 190,000 books and has already surfaced in copyright lawsuits involving Apple, Salesforce and other tech firms.

Lyon argues that because SlimPajama draws from Books3, Adobe effectively trained its AI on copyrighted works without authorization. The complaint alleges the datasets were assembled and altered in ways that violate copyright law and harmed authors whose work was scraped without consent.

Adobe has not issued a public response to the lawsuit.

A Familiar Pattern in AI Litigation

The case is far from isolated. Tech companies including Apple, Salesforce and Anthropic have all faced lawsuits over the data used to train generative AI systems. Just months ago, Anthropic agreed to pay $1.5 billion to settle claims that pirated works were used to train its chatbot, Claude.

At the center of these disputes is a common tension: AI systems require enormous volumes of text to function effectively. In the rush to build more capable models, companies often turned to large, publicly available datasets that included everything from encyclopedia entries to complete books — sometimes without confirming whether that content was properly licensed.

What once seemed like an abstract ethical concern has increasingly become a courtroom issue, with authors and publishers demanding accountability.

Why the Case Resonates Beyond the Courtroom

For marketers, publishers and content creators who rely on generative AI to speed up campaigns or automate workflows, the lawsuit raises broader concerns about legal exposure and reputational risk.

The complaint underscores a growing question: if an AI tool was trained on infringing material, who bears responsibility for the downstream use of its outputs?

Key Issues Emerging for AI Users

As lawsuits mount, experts say organizations using AI tools should pay closer attention to how those systems are trained.

Understanding data sources: Companies are increasingly expected to ask vendors how their models were built and whether licensed or human-curated data was used.

Reviewing AI-driven workflows: Documenting where and how AI tools generate content can help organizations respond quickly if questions arise about content origins.

Contract protections: Legal analysts say contracts with AI providers are likely to include stronger indemnity clauses to protect users if training data is later found to be infringing.

Responsible use policies: Clear internal guidelines on transparency, attribution and human oversight are becoming a baseline expectation rather than a best practice.

A Legal Storm Still Building

Adobe’s lawsuit is the latest sign that courts are becoming the battleground for defining how far AI developers can go when sourcing training data. As generative AI becomes embedded across industries, pressure is mounting for companies to open the black box and prove their models were built on lawful foundations.

For now, the Adobe AI pirated book suit stands as another marker in an escalating conflict — one that could reshape how artificial intelligence is trained, regulated and trusted in the years ahead.