Archive — US — techcrunch
Adobe hit with proposed class-action, accused of misusing authors’ work in AItraining
Adobe faces a proposed class-action lawsuit from author Elizabeth Lyon, alleging it improperly used pirated books, including her own, to train its AI model SlimLM. The suit claims SlimLM was developed using datasets derived from the controversial RedPajama and Books3 collections, which have sparked multiple legal challenges related to copyright infringement in AI training.
Key points
- Adobe launched AI services like Firefly, embracing AI technologies.
- A lawsuit claims Adobe's SlimLM used pirated books for AI training, particularly Lyon's work.
- SlimLM is identified as a derivative of the pre-training dataset SlimPajama, which in turn comes from the RedPajama dataset, known for containing copyrighted works.
- This lawsuit adds to a growing trend where tech companies face legal actions for using copyrighted materials in AI training datasets.
Why it matters
- The lawsuit reflects a broader trend in 2025 wherein many tech firms are facing legal issues over copyright infringement related to AI.
- Previous cases, such as Anthropic's $1.5 billion settlement with authors, illustrate the urgency in addressing legal frameworks around AI and copyright.
- Similar lawsuits against other companies like Apple and Salesforce indicate a rising concern and scrutiny over AI training practices.
The Adobe lawsuit highlights ongoing legal challenges faced by the AI industry regarding the ethical use of copyrighted material, reflecting a significant tension between innovation and intellectual property rights that is likely to shape future AI development.