Patenting AI: Navigating the Technical Maze in Europe and the US

In a recent episode of the CEIPI IP Business Talks, host Prof. Dr. Alexander Wurzer welcomed patent attorneys Robert Plotkin and Bastian Best for an insightful discussion on patenting AI innovations. This live interview, spanning three time zones, delved into the complexities of securing AI patents in Europe and the United States, highlighting the challenges and opportunities in this rapidly evolving field.

The Rise of AI Patents: A Global Perspective

As artificial intelligence continues to reshape industries, the patent landscape is evolving to accommodate these groundbreaking innovations. Both Plotkin and Best, with their extensive backgrounds in computer science and patent law, offered unique insights into how AI is transforming the intellectual property realm.

Bastian Best, a European patent attorney based in Singapore, recounted his journey into AI patents, which began with software patents about 17 years ago. He noted that the AI patent boom started around 5-6 years ago, as clients shifted from building physical prototypes to discussing neural networks, cloud training, and digital twins. Best emphasized the excitement surrounding AI patents due to the constant flux in legal frameworks and technological advancements.

Robert Plotkin, a US patent attorney with over 25 years of experience, shared his early fascination with AI, dating back to his high school days when he programmed a rudimentary machine learning system for Tic-Tac-Toe. Plotkin’s book “The Genie in the Machine,” published in 2009, explored computer-automated inventing, foreshadowing the current AI revolution in patent law.

Patenting AI – IP Business Talk with Bastian Best and Robert Plotkin

Comparing European and US Approaches to AI Patents

A significant portion of the discussion focused on the differences and similarities between European and US patent systems when it comes to AI innovations. Both experts agreed that while the terminology might differ, there is a growing alignment in the fundamental approach to AI patentability.

The European Perspective

In Europe, the concept of “technical effect” or “technical contribution” is central to patent eligibility. Best explained that the European Patent Office (EPO) consistently looks for hard technical effects, such as improved efficiency in computer resources, reduced network bandwidth, or enhanced security features. He noted that EPO examiners tend to be more consistent in their objections and conclusions compared to their US counterparts.

Best highlighted that in Europe, once an invention overcomes the technical effect hurdle, it must then pass the non-obviousness test. He emphasized the importance of focusing on tangible technical advantages when framing AI patent applications for the European market.

The US Approach

Plotkin described the US patent system’s evolving stance on AI patents. While US law doesn’t explicitly use terms like “technical effect,” the US Patent and Trademark Office (USPTO) increasingly looks for technical features within claims to satisfy patent eligibility requirements. Plotkin mentioned that inventions working “in the guts of AI,” such as new training methods for models, are generally considered patent-eligible in the US.

However, Plotkin pointed out a recent inconsistency in USPTO guidelines regarding AI-related inventions, particularly concerning the patentability of neural network training methods. This has led to increased variability in examiner decisions, making it crucial for patent applicants to prepare comprehensive applications with multiple alternatives to navigate the examination process successfully.

Key Challenges in Patenting AI Inventions

The discussion highlighted several challenges faced by inventors and patent attorneys in the AI space:

• Defining Technicality: The concept of technical contribution in AI inventions remains a challenge for both European and US patent systems. Europe generally requires “hard” technical effects, such as improved efficiency or reduced resource consumption. The US approach appears to be somewhat broader, considering practical applications as potentially sufficient for patent eligibility.
• Examiner Variability: In the United States, there is significant variation among patent examiners in their approach to AI patents. This inconsistency makes it difficult for applicants to predict examination outcomes. As a result, patent attorneys must employ flexible drafting strategies to accommodate different examiner interpretations.
• Rapid Technological Evolution: The fast-paced development of AI technology poses a significant challenge for patent systems worldwide. Patent offices struggle to keep their guidelines and examination practices up-to-date with the latest AI advancements. This constant evolution necessitates frequent revisions and adaptations in patent examination procedures.
• Detectability and Enforceability: The experts highlighted concerns about the detectability of AI patent infringement, particularly for training methods and internal processes. Balancing the need for patent protection with the practical challenges of enforcement is crucial. Patent applicants must consider the potential market value of their innovation against the difficulty of detecting infringement when deciding on their IP strategy.
• Inventorship Questions: The emergence of AI-assisted and AI-generated inventions has sparked debates about the concept of inventorship in patent law. Current patent systems do not recognize AI as an inventor, requiring human involvement in the inventive process. Both experts acknowledged that this issue might become more significant in the future as AI capabilities continue to advance.

Strategies for Successful AI Patent Applications

The experts offered several strategies for improving the chances of securing AI patents:

• Focus on Technical Effects: When patenting AI innovations, it’s crucial to emphasize concrete technical advantages that the invention provides. These could include improved efficiency in computer resource usage, reduced network bandwidth consumption, or enhanced security features. By highlighting these tangible technical benefits, applicants can strengthen their case for patent eligibility, especially in jurisdictions that prioritize “hard” technical effects.
• Detailed Specifications: Patent applications for AI innovations should include comprehensive specifications with multiple alternatives to provide flexibility during the prosecution process. This approach allows applicants to adapt their claims in response to examiner objections or changing legal interpretations. By offering a range of embodiments and implementation details, inventors can increase their chances of securing patent protection for their AI-related inventions.
• Highlight Unique Contributions: It’s essential to clearly articulate why the AI invention couldn’t be achieved using off-the-shelf solutions or through trivial adaptations of existing technologies. This involves explaining the specific challenges overcome and the novel approaches developed to address them. By emphasizing the unique contributions of the invention, applicants can better demonstrate non-obviousness and inventive step, key requirements for patentability.
• Consider the Application Domain: AI innovations applied in inherently technical fields, such as industrial machinery or medical devices, may face fewer hurdles in demonstrating technical effect. The application domain can provide a natural context for establishing the technical nature of the invention. Inventors working in these areas should leverage the inherent technicality of their field to strengthen their patent applications.
• Balance Patent Protection and Trade Secrets: For AI innovations that are difficult to detect or enforce through patents, it may be prudent to consider keeping them as trade secrets instead. This is particularly relevant for internal processes or training methods that are not easily observable by competitors. Inventors should carefully weigh the benefits of patent protection against the risks of disclosure and the challenges of enforcement when deciding on their intellectual property strategy.
• Adapt to Jurisdictional Differences: Given the variations in how different patent offices interpret “technicality” and assess AI inventions, it’s crucial to tailor patent applications to meet the specific requirements of each jurisdiction. This may involve emphasizing different aspects of the invention or using jurisdiction-specific language and claim structures. By adapting the application to local requirements, inventors can maximize their chances of securing patent protection across multiple regions.

The Future of AI Patents

Both Plotkin and Best expressed excitement about the future of AI patents while acknowledging the challenges ahead. They anticipate continued evolution in patent laws and examination practices to keep pace with AI advancements.

The experts also touched on the broader implications of AI on the patent system, including potential shifts in how we conceptualize invention and innovation. Plotkin suggested that as AI systems become more capable of generating inventions, we may need to reevaluate the fundamental principles of intellectual property protection.

Conclusion: Navigating the AI Patent Landscape

As AI continues to transform industries and push the boundaries of innovation, the patent landscape must adapt to accommodate these advancements. Patent attorneys, inventors, and businesses operating in the AI space need to stay informed about the evolving legal frameworks in different jurisdictions and develop strategies to protect their innovations effectively.

The insights shared by Robert Plotkin and Bastian Best in this SAP IP Business Talk highlight the complex interplay between technological advancement and legal frameworks in the realm of AI patents. As the field continues to evolve, close collaboration between technologists, legal experts, and policymakers will be crucial in shaping a patent system that fosters innovation while providing adequate protection for AI-driven inventions.

For those looking to deepen their understanding of AI patenting in Europe, the upcoming university certificate program on patenting AI, scheduled for April 1st and 2nd, offers an excellent opportunity to gain practical insights and navigate the complexities of this rapidly evolving field.

As we move forward in this AI-driven era, the ability to secure and leverage AI patents will likely become a critical factor in business success across various industries. By staying informed about the latest developments in AI patent law and adopting strategic approaches to patent filing, innovators can better position themselves to thrive in this dynamic landscape.

📌 If you want to learn more about AI-generated inventions, you can find the 🔎𝗜𝗣 𝗠𝗮𝗻𝗮𝗴𝗲𝗺𝗲𝗻𝘁 𝗚𝗹𝗼𝘀𝘀𝗮𝗿𝘆 entry: here

📌  If you want to learn more about software patents, you can find the 🔎𝗜𝗣 𝗠𝗮𝗻𝗮𝗴𝗲𝗺𝗲𝗻𝘁 𝗚𝗹𝗼𝘀𝘀𝗮𝗿𝘆 entry: here

📌 If you are interested in the latest developments around AI-generated inventions and patenting AI, you can find the topic page on the digital IP lexicon 🔗𝗱𝗜𝗣𝗹𝗲𝘅: here

📌 From the CEIPI IP Business Academy – European Patent Office joint distance learning University Diploma IP Business Administration we have a lecture summary with industry perspective on patenting AI on the 📑𝗜𝗣 𝗠𝗮𝗻𝗮𝗴𝗲𝗺𝗲𝗻𝘁 𝗟𝗲𝘁𝘁𝗲𝗿𝘀: here

About the interviewees

Bastian Best started his career with a degree in computer science before he joined the German patent law firm Bardehle Pagenberg. In May 2021 he founded his own patent law firm, BESTPATENT, where he helps companies make sense of patents in the digital transformation. His special interest lies in the patentability issues involved in artificial intelligence, machine learning, blockchain, cryptography and the Internet of Things.

Robert Plotkin is an MIT-educated computer scientist and engineer with a technical background matched by few other attorneys. In his two decades as a patent attorney specializing in software patents, Mr. Plotkin has obtained hundreds of patents for leading technology companies worldwide, including HP, Compuware, Tata Consultancy Services, and 3M.