Renowned author sues Adobe for pirating authors' books to train its AI model SlimLM

In an unfortunate occurrence in its strides to integrate artificial intelligence (AI) into its design tools, Adobe has been sued with a class-action lawsuit filed by author Elizabeth Lyon, claiming the company pirated her books to train its AI model, SlimLM.

The lawsuit by Lyon alleges that the globally acclaimed computer software company incorporated unauthorised works into its training dataset, specifically the SlimPajama dataset, part of the RedPajama dataset known for containing copyrighted material.

Who is Elizabeth Lyon, who has sued Adobe for misusing authored works for AI training?

For those unfamiliar, Elizabeth Lyon is a widely celebrated writing teacher and book editor since 1988, famous for her booklet series on how to write, edit and market novels and non-fiction.

Lyon claims that her writings, along with those of other authors, were included in the abovementioned dataset, which Adobe used to develop SlimLM, a small language model designed for document assistance on mobile devices.

The lawsuit underlines that SlimPajama was created by copying and altering the RedPajama dataset, which includes a collection known as Books3, consisting of 191,000 books.

The legal action forms part of a recently emerged trend in the tech space where tech giants are facing lawsuits for allegedly using copyrighted materials without permission to train AI systems.

Similar petitions have been filed, including a lawsuit against Apple for using copyrighted content in its AI model and another against Salesforce for employing the same RedPajama dataset.

The ongoing legal challenges hint at the complexities around AI training data and copyright laws. In a surprising case, Anthropic recently agreed to pay $1.5 billion to authors who accused it of using pirated works for its popular chatbot Claude.

These happenings signify the increasing scrutiny on how AI algorithms are developed and the materials they're based on.