Anthropic sued over neural network patents in AI legal clash

The University of Tennessee Research Foundation sued Anthropic in Delaware federal court over patents it says cover neural-network technology inspired by neuroscience, opening another legal front for one of the best-known builders of large language models.

The complaint, described by multiple outlets as involving two patents, lands in a market where the biggest AI firms are already fighting over copyright, data use and the commercial value of the techniques behind generative systems. If the court sides with the plaintiff, the remedies could include damages, injunctions or a forced licensing deal, outcomes that could raise costs not only for Anthropic but for other companies competing with OpenAI and Google in the race to ship more capable models.

The plaintiff is the University of Tennessee Research Foundation, a nonprofit that licenses the university’s intellectual property. One outlet described the filing as the first patent infringement case against Anthropic, underscoring how quickly disputes over neural-network building blocks are moving from academic history to active litigation. The patents at issue are tied to neuroscience-inspired neural networks, a reminder that many of the ideas now powering commercial AI systems were developed first in university labs before being folded into products sold at scale.

The University of Tennessee’s research arm has signaled for years that it was building a patent portfolio around this kind of work. In a July 24, 2023 post, UTRF said the TENNLab Research Group had been issued seven patents and had four patents pending in neuromorphic computing research. That kind of portfolio is exactly what can turn foundational research into leverage when AI companies begin paying billions of dollars to train, deploy and defend their models.

For Anthropic, the case arrives as the company tries to position itself as a safety-conscious alternative to its larger rivals. That branding has not insulated it from the widening legal fight around AI development. A U.S. judge approved Anthropic’s $1.5 billion copyright settlement on July 20, a day before the patent suit became public, making the company’s courtroom exposure look broader and more expensive by the day.

The new case also tests whether courts will treat neural-network techniques as general engineering methods or as protected inventions that must be licensed before they can be used in foundation models. For an industry racing to scale, that question goes directly to who controls the core machinery of AI, and how much it will cost to build the next generation of systems.