Anthropic and OpenAI Sued Over Infringement of Neural Network Patents
Anthropic and OpenAI face patent infringement lawsuits from Flinn AI Technologies over five neural network patents allegedly used in their AI systems.
The artificial intelligence industry is facing another significant legal challenge as two of its biggest players, Anthropic and OpenAI, have been sued over allegations that their technologies infringe patented neural network inventions. The lawsuits, filed in federal court in Delaware, claim that both companies have incorporated patented machine-learning technologies into their artificial intelligence systems without obtaining the necessary authorization or licensing agreements.
Flinn Artificial Intelligence (AI) Technologies LLC, the company bringing the lawsuits, alleges that Anthropic’s Claude and OpenAI’s ChatGPT rely on technologies covered by five patents developed by computer scientists Steven Flinn and Naomi Moneypenny. The cases introduce another potential intellectual property dispute for an industry already facing increased scrutiny over how AI companies develop, train, and commercialize their products.
Anthropic and OpenAI Face Patent Infringement Lawsuits
On November 12, 2026, Flinn Artificial Intelligence Technologies filed two separate complaints in the U.S. District Court for the District of Delaware. One lawsuit names Anthropic PBC as the defendant, while the other targets OpenAI OpCo LLC.
According to the complaints, the five patents involve attention-based neural network architectures, a foundational technology used in modern artificial intelligence systems. Flinn claims that the inventions provide essential capabilities that allow AI models to process information, retrieve relevant data, generate responses, and perform increasingly sophisticated tasks.
The lawsuits identify the following patents:
- U.S. Patent No. 12,223,404
- U.S. Patent No. 12,327,166
- U.S. Patent No. 12,462,197
- U.S. Patent No. 12,462,198
- U.S. Patent No. 12,511,580
Steven Flinn and Naomi Moneypenny reportedly filed the underlying machine-learning patent applications in 2015 and 2016 before assigning their rights to Flinn Artificial Intelligence Technologies.
Flinn argues that Anthropic and OpenAI have built substantial businesses around capabilities described in these inventions without compensating the patent owners.
Why Attention-Based Neural Networks Matter to Anthropic and OpenAI
To understand the significance of these lawsuits, it helps to look at the technology involved.
Neural networks are computational systems designed to identify patterns, process information, and perform tasks that traditionally require human intelligence. Modern large language models use sophisticated neural network architectures to understand relationships between words, interpret context, and generate relevant responses.
One particularly important development is attention-based processing, which allows an AI model to determine which pieces of information are most relevant when interpreting or generating content.
For example, when someone asks ChatGPT a complicated question, the system must evaluate the relationship between different words, previous information, and the overall context before generating an answer.
The patents identified in Flinn’s complaints allegedly cover iterative attention-based neural network training and processing. These architectures describe multiple stages of processing in which attention mechanisms, representations, and generated probabilities interact to influence subsequent operations.
According to Flinn, the patented inventions relate to several capabilities associated with today’s advanced AI products, including:
- Reasoning and inference: Processing information through multiple stages to generate more sophisticated responses.
- Memory and personalization: Using previously processed information to improve future interactions.
- Retrieval and search: Identifying relevant information and incorporating it into generated responses.
- Agentic workflows: Enabling AI systems to perform tasks through coordinated operations.
- Coding capabilities: Processing programming instructions and assisting with software development.
- Multimodal processing: Working with different forms of information, including text, images, and other media.
The complaint also references Anthropic’s May 2024 interpretability research involving attention heads in its Claude 3 Sonnet model as part of its allegations.
However, the presence of similar technical concepts does not independently establish patent infringement. The court will ultimately need to evaluate the specific patent claims and the accused systems.
What Damages Are Anthropic and OpenAI Facing?
Flinn is seeking monetary damages and additional legal remedies in both cases. The complaint against Anthropic specifically alleges willful infringement, potentially opening the door to enhanced damages if the allegations are established.
Among the requested remedies are:
- Monetary compensation for past alleged infringement.
- Reasonable royalties for continued use of the patented technologies.
- Enhanced damages of up to three times the amount awarded, if legally justified.
- Recovery of attorney fees and litigation expenses.
- Prejudgment and post-judgment interest.
- An accounting of profits or other financial relief deemed appropriate by the court.
The company has not publicly established a specific dollar amount for the damages it seeks.
Given the commercial scale of both AI businesses, the financial implications could become substantial if Flinn ultimately succeeds. However, the amount of any potential award would depend on the evidence, applicable patent claims, and the court’s determination.
A New Legal Challenge for the Artificial Intelligence Industry Especially – Anthropic and OpenAI
Patent infringement litigation represents another growing concern for companies developing commercial AI technologies. While lawsuits involving copyrighted books, articles, images, and other training materials have received considerable attention, disputes involving the underlying technical architecture of AI systems introduce a different category of intellectual property risk.
For companies like Anthropic and OpenAI, the outcome could have implications beyond the immediate financial damages. A successful patent challenge could potentially influence licensing arrangements, technology development strategies, and how certain AI capabilities are implemented in future products.
For now, the allegations remain unproven, and neither lawsuit has established that either company violated the patents.
As artificial intelligence continues to evolve, the dispute highlights an increasingly important question: Who owns the underlying innovations that make today’s most advanced AI systems possible, and what obligations do developers have when commercializing those technologies?