USPTO's Updated Guidance on AI Inventions

The United States Patent and Trademark Office (USPTO) has released updated guidance on patent subject matter eligibility. This update focuses on critical and emerging technologies, particularly artificial intelligence (AI). The new guidance aims to assist USPTO personnel and stakeholders in evaluating the subject matter eligibility of claims in patent applications and patents related to AI technology and inventions. Furthermore, it introduces new examples to help apply the USPTO's subject matter eligibility guidance during patent examination, appeal, and post-grant proceedings. The guidance also addresses stakeholder feedback and recent Federal Circuit decisions.

USPTO's Updated Guidance on AI Inventions

Background

Recognizing the potential of responsible AI use to solve urgent challenges and foster innovation, President Biden issued Executive Order 14110. This Executive Order emphasizes the importance of promoting innovation, competition, and collaboration to enable the United States to lead in AI and leverage its potential to tackle societal challenges. It underscores the necessity of investing in AI-related education, training, development, research, and capacity while addressing intellectual property issues to protect inventors and creators.

In accordance with Executive Order 14110, the USPTO has issued prior guidance on AI topics and issues, and most recently issued updated guidance on patent subject matter eligibility specifically for AI inventions. The guidance update clarifies how AI inventions fit within the four categories of patent-eligible subject matter under 35 U.S.C. 101: processes, machines, manufactures, and compositions of matter. The courts have deemed abstract ideas, laws of nature, and natural phenomena (including products of nature) as exceptions to patent eligibility. This update assists USPTO personnel and stakeholders in evaluating AI inventions' eligibility, provides background on the USPTO's efforts related to AI, and addresses stakeholder feedback.

Updates Provided by the USPTO

Artificial Intelligence Efforts

Since August 2019, the USPTO has actively sought public input on patenting AI inventions. Following a comprehensive review of stakeholder feedback, the USPTO published a report in October 2020 titled "Public Views on Artificial Intelligence and Intellectual Property Policy." The report highlighted that most stakeholders viewed AI as a subset of computer-implemented inventions, and current USPTO guidance on patent subject matter eligibility and disclosure of computer-implemented inventions was adequate to handle AI advances. However, some commenters expressed concerns that AI inventions might be at risk under the subject matter eligibility analysis due to their potential characterization as abstract ideas.

Guidance Update and Impact on Examination Procedure

The new guidance update provides an overview of the existing patent subject matter eligibility guidance and discusses areas particularly relevant to AI inventions. It addresses whether a claim recites an abstract idea and whether it integrates the judicial exception into a practical application, such as improving the functioning of a computer or another technology or technical field. The update also announces new examples to assist examiners and practitioners in applying the USPTO's subject matter eligibility guidance to AI inventions during the patent examination process.

Clarifying Subject Matter Eligibility

One key aspect of the updated guidance is the clarification on how AI inventions are analyzed for subject matter eligibility. The guidance emphasizes that AI inventions, often characterized as computer-implemented inventions, must be evaluated within the framework of the existing subject matter eligibility analysis. This involves determining whether a claim recites a judicial exception (such as an abstract idea) and whether the claim as a whole integrates the exception into a practical application.

The USPTO provides examples to illustrate how this analysis applies to AI inventions. For instance, claims that involve specific applications of AI to solve technical problems or improve technological processes are likely to be deemed eligible. On the other hand, claims that merely recite abstract ideas without sufficient additional elements to integrate the exception into a practical application may be considered ineligible.

Addressing AI-Assisted Inventions

The guidance update also addresses AI-assisted inventions, which are creations developed by natural persons using AI systems. The USPTO clarifies that the method of developing an invention, including the use of AI, does not impact the subject matter eligibility analysis under 35 U.S.C. 101. Instead, the focus is on the claimed invention itself and whether it is the type of innovation eligible for patenting.

In addition to subject matter eligibility, the USPTO has issued guidance on inventorship for AI-assisted inventions. According to current statutes, contributions by tools such as AI systems do not qualify for inventorship recognition. However, inventions created with significant human contributions, even if AI-assisted, can still be eligible for patent protection.

Examples and Practical Applications

To aid in the practical application of the updated guidance, the USPTO has also developed new subject matter eligibility examples for AI inventions. These examples provide hypothetical claim scenarios and analyses to illustrate how the eligibility criteria are applied.

Example 47: AI for Anomaly Detection

One example involves using an artificial neural network to identify or detect anomalies. The example demonstrates how claims that specify the technical implementation and application of AI for anomaly detection can be analyzed to determine subject matter eligibility. By focusing on the specific technological improvements provided by the AI system, the example helps clarify how such claims can be integrated into a practical application.

Example 48: AI in Speech Signal Analysis

Another example involves AI-based methods for analyzing speech signals and separating desired speech from background noise. This scenario highlights how AI technologies can improve existing processes and provide practical solutions to technical problems. The example shows how claims that describe the specific steps and technical details of the AI-based method can be evaluated for subject matter eligibility.

Example 49: Personalized Medical Treatment

A third example involves an AI model designed to assist in personalizing medical treatment based on individual patient characteristics. This example illustrates how AI can be applied to enhance healthcare outcomes by tailoring treatments to specific patient needs. The example provides insights into how claims related to AI-driven medical innovations can be analyzed for subject matter eligibility.

Conclusion

The USPTO's updated guidance on AI inventions marks a significant step in clarifying the patent subject matter eligibility criteria for AI technologies. By incorporating stakeholder feedback and recent court decisions, the guidance aims to promote innovation and provide clarity to both USPTO personnel and applicants. The introduction of new examples further enhances the practical understanding of how AI inventions can be evaluated for patent eligibility.

As AI continues to evolve and play a transformative role across various industries, the USPTO's efforts to provide clear and consistent guidance will be crucial in fostering innovation and ensuring that the United States remains at the forefront of AI advancements.

AI for patents.

Be 50%+ more productive. Join thousands of legal professionals around the world using Solve’s Patent Copilot™ for drafting, prosecution, invention harvesting, and more.

Related articles

How Much of Your Patent Practice Should You Codify?

Both in-house teams and outside counsel can let a purpose-built platform carry the shared foundation for their patent work. They can then focus their limited time on the standards and judgment that set their work apart. AI has made this division of labor more valuable by raising the payoff for turning a practice into templates, instructions, and review criteria that run at scale. Let’s call that codification. The platform can encode a great deal of best practice out of the box, and a team can add its own custom templates and instructions on top. The real question is how much to add and how much to leave to the platform. There is no fixed formula because the right balance changes as the team’s practice, the law, and the technology evolve.

Key takeaways

• AI has raised the payoff for codifying patent practice. Codification now guides AI-assisted drafting and review directly, not just junior training.

• A purpose-built patent platform can encode a large body of best practice out of the box, so a team can begin with that foundation rather than write its own.

• Solve Intelligence maintains its shared foundation across more than 700 IP teams, keeping it current far more efficiently than any single team could alone.

• The strategic choice is how much of your own practice to codify on top; scarce expert time should go to the standards and judgment that differentiate your work.

• A named professional still signs off on every filing, so accountability for the result never shifts to the AI.

How Successful Patent Practitioners Are Putting AI to Work

The most effective patent practitioners are already using AI patent drafting to draft faster, catch claim inconsistencies earlier, and free up hours for the strategic work that actually wins allowances.

Key takeaways

  • AI patent drafting tools can reduce application drafting time by up to 80 percent, with Solve Intelligence customers consistently reporting 50 percent or greater efficiency gains across drafting and prosecution 
  • Roughly 9 out of 10 utility patent applications receive at least one office action rejection, so prosecution efficiency matters as much as drafting speed 
  • Solo attorneys use AI to match larger law firms on turnaround speed and client capacity
  • The strongest reported results come from iterative AI–attorney collaboration, with practitioners directing the process and owning the final work product

Adopting AI in Patent Work: A Practical Playbook for IP Teams

Solve Intelligence works with over 700+ patent teams as they bring AI into daily practice, and the same pattern shows up again and again: recognising that AI helps is easy, but building consistent, team-wide use is not. Adoption tends to stall for a handful of reasons, from informal early experiments to unclear decision making, and a promising trial can fade out without anyone establishing whether the tool or the rollout was at fault. This playbook lays out the process that gets a team from first experiment to settled habit, with the attorney's judgment in control at every step.

How AI Brings Patent Intelligence Into Every Decision

AI makes it practical to rerun patent intelligence as products and patent rights develop. For example, at concept stage, broad freedom-to-operate screening identifies the rights that merit attention. As the design matures, selected patents are escalated for feature-by-feature claim charts, while scheduled monitoring refreshes the analysis when claims are amended or an application proceeds to grant.

This contrasts with the traditional approach, in which landscapes, FTO reviews, and portfolio analyses were commissioned as separate projects at fixed stages. Each took substantial time to complete and was rarely repeated.