Can I patent AI inventions - a short guide

Artificial intelligence (AI) has become a cornerstone of technological innovation, and the United States Patent and Trademark Office (USPTO) has recognized AI-driven inventions as patentable, provided they meet established criteria such as novelty, non-obviousness, utility, and patent eligibility. Early AI patents date back to the 1960s, involving neural networks and audio pattern recognition technologies. Today, the patent landscape for AI spans groundbreaking algorithms, hardware innovations, and applications, showcasing the dynamic interplay between AI and intellectual property law.

Patent Criteria and Human Inventorship

The USPTO mandates that only natural persons can be listed as inventors on patents. This principle was reinforced in Thaler v. Vidal (2022), where the Federal Circuit ruled that inventions solely created by AI systems are ineligible for patents. However, AI can collaborate with humans in the inventive process. For AI-assisted inventions to qualify, human inventors must significantly contribute to the invention’s conception, demonstrating ingenuity that transcends merely using AI tools.

The USPTO has provided detailed guidelines for determining inventorship in AI-assisted innovations, incorporating criteria like the Pannu factors, which assess the significance and quality of human contributions. These guidelines ensure that human ingenuity remains central to patentable inventions, even as AI is increasingly supportive.

Technologies in AI Patents

AI patents cover a wide range of technologies, including:

1. Generative AI and Large Language Models (LLMs): Patents in this area involve training methods and architectures for generating text, images, audio, and video. These innovations improve efficiency, scalability, and the realism of AI outputs.
2. Generative Adversarial Networks (GANs): GAN patents focus on novel architectures and techniques that enhance training stability, reduce issues like mode collapse, and expand applications in medical imaging, entertainment, and more.
3. Reinforcement Learning (RL): Patents in RL target systems enabling intelligent decision-making in robotics, gaming, and autonomous systems.
4. Neural Symbolic Computing: This hybrid approach combines neural networks with symbolic reasoning, allowing AI to integrate learning and logical inference. Patents cover algorithms and architectures that advance this fusion.
5. AI Hardware and Knowledge Processing: AI-specific hardware, such as Google’s Tensor Processing Unit (TPU), has enabled efficient neural network computations, while knowledge processing patents address methods for deriving insights from structured data.

Patent Eligibility Under Section 101

AI-related inventions must adhere to Section 101 of the U.S. Patent Act, which outlines patent eligibility criteria. Following the Supreme Court’s Alice Corporation decision (2014), determining eligibility for software-related inventions, including AI, has become nuanced. Patent applications must demonstrate that the AI innovation is not abstract but constitutes a tangible application. The USPTO has issued specific examples, such as methods for training neural networks, to guide inventors through Section 101 considerations.

AI-Assisted Inventions and Inventorship

With the growing role of AI in the inventive process, the USPTO has clarified the standards for patenting AI-assisted inventions. The five guiding principles emphasize:

1. AI use does not diminish human contributions.
2. More than merely presenting a problem to AI is required for inventorship.
3. Significant contributions to AI output qualify as inventorship.
4. Contributions to foundational elements of the invention can establish inventorship.
5. Ownership or oversight of AI systems equates to inventorship with meaningful input.

These principles aim to balance the supportive role of AI with the centrality of human creativity, ensuring the patent system adapts to the evolving technological landscape.

Conclusion

The patentability of AI inventions reflects the increasing importance of AI in innovation while maintaining a legal framework that prioritizes human ingenuity. As AI capabilities expand, U.S. patent law continues to evolve, addressing the complexities of AI-driven and AI-assisted inventions to foster innovation while upholding the principles of intellectual property rights.