Adobe Under Fire for Alleged AI Copyright Violations
Adobe’s aggressive push into AI could be hitting a legal roadblock. The software giant, known for creative tools like Photoshop and Illustrator, is now facing a proposed class-action lawsuit claiming it used copyrighted books to train its AI systems. Authors are arguing that Adobe incorporated their works without permission, raising fresh questions about intellectual property in the AI era.
Credit: Jaque Silva/SOPA Images/LightRocket / Getty ImagesThis lawsuit highlights growing concerns over how AI models are trained. Many tech companies rely on large datasets, often scraped from publicly available content. For authors, this practice can feel like their creative work is being exploited without recognition or compensation.
Lawsuit Details: Author Elizabeth Lyon Takes the Lead
The class-action, filed on behalf of Oregon author Elizabeth Lyon, alleges that Adobe used pirated copies of books—including Lyon’s own publications—to train its SlimLM AI model. SlimLM is marketed as a small language model optimized for document assistance on mobile devices.
Adobe claims SlimLM was pre-trained on SlimPajama-627B, an “open-source, multi-corpora dataset” released by Cerebras in June 2023. Lyon argues, however, that SlimPajama is essentially a derivative of the controversial RedPajama dataset, which includes Books3—a massive collection of 191,000 copyrighted books.
The Controversial Datasets: RedPajama and Books3
Books3 and RedPajama have been central to legal disputes in AI development. These datasets have been widely used to train generative AI models, but they contain copyrighted works that were not always licensed for such use.
According to Lyon’s complaint, Adobe’s SlimLM relied on a processed subset of SlimPajama that contained her writings. The lawsuit claims, “SlimPajama contains the Books3 dataset, including the copyrighted works of Plaintiff and the Class members,” suggesting widespread use of unlicensed material.
Adobe’s AI Expansion Faces Legal Challenges
Adobe has invested heavily in AI since 2023, rolling out tools like Firefly, its AI-powered creative suite. While these services have been embraced by designers and marketers, the legal risks of training AI on copyrighted content are becoming harder to ignore.
This isn’t the first time major tech companies have faced similar accusations. In September, Apple was sued for allegedly using copyrighted material from Books3 to train its Apple Intelligence AI. These cases are setting important precedents for how AI developers can source training data legally.
Authors Push Back Against AI Misuse
For authors like Lyon, the lawsuit represents a fight for recognition and compensation. Many writers feel that AI companies profit from their creative output without offering any form of payment or credit.
Legal experts note that these disputes could reshape the AI landscape, forcing companies to obtain licenses or create new datasets that avoid copyrighted material. Until courts make definitive rulings, authors and AI developers will continue to clash over ownership and fair use.
SlimLM and Adobe’s Defense
Adobe maintains that SlimLM was built on open-source datasets designed to avoid copyright infringement. The company describes SlimLM as a mobile-optimized model for document assistance, emphasizing that it was pre-trained on publicly available data.
Yet, Lyon’s lawsuit argues that the distinction between open-source and derivative datasets is legally murky. By incorporating SlimPajama, which allegedly included Books3 content, Adobe may have crossed a line.
The Wider Implications for AI Development
This case underscores the broader tension between innovation and copyright law. Generative AI systems rely on enormous datasets to function effectively, but not all of those datasets are free from legal challenges.
Tech companies face growing scrutiny over transparency, licensing, and ethical AI training. Lawsuits like Lyon’s could prompt stricter regulations, potentially changing how AI is developed and marketed in the future.
Legal Experts Weigh In
Intellectual property attorneys warn that Adobe’s case could influence how courts interpret copyright infringement in AI contexts. If Lyon’s lawsuit succeeds, companies may need to rethink their training methods or invest in new, fully licensed datasets.
Some analysts predict that more authors and creators will follow Lyon’s lead, resulting in a wave of legal actions against AI developers. The stakes are high—not just for Adobe, but for the entire generative AI industry.
What This Means for Authors and Creators
For authors, this lawsuit is a signal that their rights cannot be ignored in the AI era. Many creators are calling for fair compensation models and clear rules for AI training. The outcome of Adobe’s case could set important legal precedents that shape the future of creative industries.
The case also emphasizes the need for transparency from AI developers. Users and creators alike are demanding clearer information about where AI models source their data, ensuring ethical use of intellectual property.
Adobe and the AI Industry
Adobe’s legal battle highlights a crucial moment for AI regulation. Companies are innovating faster than laws can adapt, and courts may now play a decisive role in defining the limits of AI training.
As AI tools like SlimLM become more widespread, the tension between creativity, copyright, and technological progress will only grow. The Adobe lawsuit could mark the beginning of a new era in AI accountability.
The Future of AI and Copyright
Adobe’s proposed class-action lawsuit brings attention to the complex relationship between AI innovation and copyright law. Authors like Elizabeth Lyon are challenging the tech industry, demanding recognition for their work.
The outcome could influence how AI companies source training data, potentially shaping the next generation of AI models. For creators, it’s a pivotal moment to assert their rights—and for the tech industry, a wake-up call about responsible innovation.
