2 min read
Adobe has leaned heavily into artificial intelligence since 2023. The software giant launched several AI services, including its “Firefly” media suite. However, this aggressive expansion has led to significant legal trouble. A recent proposed class-action lawsuit accused the software giant of using pirated books to train its AI models.
Elizabeth Lyon, an author from Oregon, filed the lawsuit. She writes guidebooks for non-fiction writing. Lyon claims that Adobe used pirated versions of her books to train its “SlimLM” program. Adobe markets SlimLM as a small language model series. It optimises document assistance tasks for mobile devices.
Adobe & The “Books3” Connection
The core of the lawsuit involves the datasets used for pre-training. Adobe stated that SlimLM was trained on SlimPajama-627B. This is an open-source dataset released by Cerebras in June 2023. However, the lawsuit alleges a “contamination” chain.
According to the filing, SlimPajama was created by copying and manipulating the RedPajama dataset. RedPajama contains “Books3”, a notorious collection of 191,000 pirated books. Consequently, the lawsuit argues that SlimPajama contains copyrighted works from Lyon and other class members. Adobe allegedly used these protected works without providing consent, credit, or compensation.
A Growing Legal Crisis for Tech Giants
Adobe is not the only company facing these accusations. “Books3” and RedPajama have caused ongoing legal trouble for the entire tech community. In September 2025, Apple faced a lawsuit for using similar copyrighted material to train Apple Intelligence. Salesforce faced a similar lawsuit in October 2025 for using RedPajama.
The financial stakes are massive. Recently, Anthropic agreed to pay $1.5 billion to settle a lawsuit with authors regarding its chatbot, Claude. That case proved that using pirated data is a multi-billion-dollar liability. This new litigation against Adobe represents a major turning point in the battle over AI training data.
