Patentability of AI Related Inventions

The rapid advancement and widespread availability of artificial intelligence (AI) tools have significantly lowered the barrier to entry for developing AI-based inventions. Open-source frameworks, cloud computing platforms, and pre-trained models enable individuals and small organizations to build sophisticated systems without the need for extensive infrastructure or deep technical expertise. As a result, innovation is no longer limited to large, well-funded entities; startups, independent developers, and even non-technical users can meaningfully contribute to the creation and commercialization of AI solutions. This democratization of AI development is accelerating the pace of innovation while also raising important questions about differentiation, patentability, and the threshold for inventive contribution.

A common question we hear is whether AI inventions are patentable and if so, what type of AI inventions? The bottom line is that the U.S. patent office welcomes AI patent applications and AI inventions are patentable if they focus on a “technical solution to a technical problem”.  AI technology is moving quickly, so it is important to also move quickly in protecting your AI innovations.  

Patent Office Leadership in AI Related Technologies

For years, courts have expanded judicial exceptions to patentable subject matter under 35 U.S.C. § 101 (abstract ideas, laws of nature, natural phenomena), creating uncertainty and high rejection and invalidation rates for many applications covering cutting edge technologies. This resulted in a trend at the United States Patent and Trademark Office (USPTO) toward excessive patent rejections based on a finding that the invention does not comprise subject matter that is patent eligible —an area that has disproportionately affected emerging technologies such as artificial intelligence (AI).

New USPTO Director John Squires (appointed September 2025) is seeking to curb these excessive rejections and is working aggressively to restore balance to a patent system where emerging technologies, including AI, have faced an uphill battle

Squires was sworn into office as the new director of the USPTO on Sept. 23, 2025. On his first full day on the job, he signed the first two patents of his tenure: one directed to distributed ledger technologies and another to medical diagnostics — both areas frequently scrutinized for patent eligibility.

At the accompanying signing ceremony,  he explained,

"I wanted to send a clear message with the first two patents issued on my watch: the U.S. Patent Office is open for business, especially for the technologies of tomorrow."

More substantively, the USPTO issued the precedential decision which overturned overly broad subject matter rejections that impacted AI inventions and emphasized a return to the traditional statutory provisions—novelty (§ 102), obviousness (§ 103), and adequate disclosure (§ 112) - as the primary tools for evaluating patentability.  In a memo to the Patent Examining Corps, Director Squires specifically noted the importance of AI inventions and cautioned examiners to heed the warning of this precedential decision against

"overbroad Section 101 rejections because '[c]ategorically excluding AI innovations from patent protection in the United States jeopardized America's leadership in [ ] critical emerging technolog[ies].'"

The takeaway is that the USPTO is open for business and welcomes AI related patent applications

AI Related Patentable Subject Matter

For now, it is settled that AI related inventions are patentable, but the AI world is vast and like other technologies, not every aspect is patentable. Generally speaking, an AI invention is patentable if the AI invention improves the functioning of a computer or another technology or technical field. The USPTO characterizes this as “a technological solution to a technological problem.” General examples of these include:

 

• Enhanced computer vision capabilities
• More efficient data processing techniques
• Increased speed or accuracy in machine learning operations
• Novel neural network architectures that solve previously unsolvable problems
• Reduced computational resource requirements
• Improved ability to handle noisy or incomplete data

In recent USPTO subject-matter eligibility guidance showing how AI inventions can be patent-eligible the USPTO provided more examples of patentable AI that involves concrete technical improvements or specific applications of AI.

Neural Network for Anomaly Detection - A system using an artificial neural network structured on a specific physical device (e.g., ASIC with defined neuron arrays) to detect anomalies in data such as security threats. This kind of claim was found eligible because it involved a specific machine and technical implementation.

AI-Based Speech Signal Processing - A trained machine learning model is used to separate desired speech from background noise. Depending on the claim’s details, this can be patentable if it’s tied to a concrete system or improves how speech is processed by computers.

AI-Assisted Personalized Medical Treatment - Using an AI model that processes patient data to generate customized treatment plans. Claims can be eligible if they include specific technical steps or particular treatment actions beyond abstract data analysis.

These examples don’t show actual issued patents but are used by the USPTO to illustrate how AI inventions can meet subject-matter requirements if they are framed around tangible improvements or specific technical processes rather than abstract ideas alone.

While not USPTO “official examples,” many AI-related inventions have been successfully patented in the U.S. because they solve real technological problems:

AI systems for conversational interfaces - For example, inventions around dialogue systems that simulate human interaction using specific architectures and response methods.

Machine learning for autonomous vehicle decision-making - Patents covering how sensory input is processed and translated into vehicle control decisions, improving safety and navigation.

AI-based fraud detection systems - Using trained models to identify suspicious patterns in transaction data to reduce false positives while improving detection rates.

AI-enhanced automation tools - Systems where AI improves performance of industrial machines, robotic control, or optimization of manufacturing workflows.

These are only some of the many examples of patentable AI subject matter. The USPTO (and other patent offices) generally look for subject matter that:

• Applies AI in a specific technical context (i.e. provides a technical solution to a technical problem)
• Improves how computers or systems operate
• Avoids overly broad or purely abstract descriptions of algorithms

Practical takeaways for drafting AI related patent applicants include:

• Draft AI and software claims to emphasize specific technical improvements.
• Provide details on the particular AI model used.
• Tie claims to practical applications rather than abstract algorithms.
• Provide measurable performance data where possible.