FORUM IP Talks: Claudia Schwarz on patent claims for CII in times of AI and machine learning

Jean-Claude: What are the unique key factors to be considered when drafting patent claims for AI and machine learning inventions compared to traditional software patents?

Claudia: Drafting patent claims for AI and machine learning involves unique challenges such as defining the appropriate claim categories and necessary definitions in the description. Unlike traditional software patents, AI-related inventions often require claims covering data pre-processing, model training, and claims covering the inference phase with the technical applications and effects of these processes. As AI and ML are typically very abstract, it is crucial to properly elaborate the main technical applications, which is typically a task for the patent attorney and rarely done by the inventor. Strategic opportunities arise from the growing importance of AI in almost all technical fields, making these patents highly valuable and strategically important for the companies.

How do you ensure that patent claims for CII in general and in particular for AI and machine learning technologies meet the sufficiency of disclosure requirements, especially given the rapid advancements in these fields?

Yes, indeed, this is one of several requirements that need to be met. When drafting such patent applications, it is usually necessary to have several discussions with the inventors to work out where it is necessary to provide more details, such as the training data and/or the specific technical implementations, and where it is not. This evaluation has to map the core idea of the invention to the requirements of recent case law. It will therefore vary from application to application. There is no "one size fits all" solution”.

What strategies do you recommend for companies to optimize their patent filing processes for CII to enhance enforceability and scope of protection?

It would be prudent for companies to adopt a cost-benefit aware approach to their CII patent filing strategy, taking into account the requirements of case law in the respective countries of interest. This includes a decision on which invention disclosures are already worth filing and which require further revisions by the inventors, for example relating to their technical contributions. This process does take some time, but may enhance the quality and enforceability of the patents. It is therefore beneficial to avoid spending money on patent drafting and filing when the chances of success for being able to draft a legally valid patent are low.

How do patent offices around the world differ in their approach to evaluating AI and machine learning patent applications, and what should applicants be aware of?

There is considerable variation between patent offices in their patentability evaluation criteria. For example, the EPO places greater emphasis on the technical aspects of an invention, whereas the USPTO tends to focus more on its practical applications. China also has a technicity requirement, though this is interpreted differently from that in Europe or Germany. It is important for applicants to be aware of these differences and to adapt their applications accordingly. From a cost-saving perspective, it is important to be aware of the specific requirements of each jurisdiction when drafting a first filing for a subsequent application, for example, a PCT application.

"...important to work together across different disciplines..."

What role do you see for interdisciplinary collaboration in the patenting process of CII technologies, and how can this enhance the quality of patent claims?

It's really important to work together across different disciplines when you're patenting something in the field of CII. That's why at our law firm all of our attorneys are qualified in CII but in addition have different technical backgrounds, like mechanical engineering, physics, electrical engineering, chemistry, etc. We work in teams and this really helps us to get a full understanding of the invention, which leads to more robust and well-defined patent claims. This collaboration ensures that all technical and legal aspects are thoroughly addressed, enhancing the overall quality of the patents.

Can you discuss the impact of recent legal developments on the patentability of CII and in particular AI and machine learning inventions and how they might influence future patent strategies?

Recent legal developments, such as e.g., the decision of the Enlarged Board of Appeal of the EPO in G1/19 and the Referral on claim interpretation in G1/24 (not yet decided), as well as the Alice framework of the US Supreme Court, have had and will have a significant impact on the practice of drafting good CII and AI applications. These developments highlight the importance of demonstrating a clear technical contribution and avoiding overly broad and abstract claims. Future patent strategies should focus on adapting to these changes by emphasizing the need for inventors to provide practical applications and concrete technical improvements of their ideas in their invention disclosures.

What best practices would you recommend for patent practitioners to stay ahead in the rapidly evolving field of AI and machine learning patents?

The world of informatics is moving fast. So, it's important for patent practitioners to keep up with the latest tech, like latest AI models. But it's also really crucial to keep an eye on case law. Training courses and industry conferences can help practitioners stay ahead of the curve in this fast-paced field.

Thank you very much for the interview, Claudia! I look forward to seeing you at our upcoming course on claims in times of AI and machine learning.

?About the interviewee:

?

Dr. Claudia Schwarz is founding partner of the patent law firm Schwarz + Kollegen Patentanw?lte, Munich. She is a computer scientist and provides worldwide patent counseling to digital technology. With her over 30 years of patent experience, she understands clients’ IP needs, and effectively protects and monetizes their most valuable R&D assets. Claudia has represented clients, including research institutions, SMEs and startup companies as well as large industry companies in matters covering a broad range of digital technologies. She has been involved and consulted in numerous patent proceedings, EPO oppositions and appeals and patent infringement matters and offers extensive experience for her clients. She is a regularly invited speaker at conferences.

Claudia had teaching positions at two universities and gives numerous seminars and lectures. Professional experience in her area of expertise is incorporated in the book titled “Computer implemented inventions – Patent protection of software?” which she wrote together with Sabine Kruspig, formerly director in computers at the EPO, and which was published as 3rd edition in German language by Heymanns’s Verlag in 2023.

She is currently working with international colleagues on a second edition of another book in English to compare the international legal situation between the main jurisdictions for software protection from a practitioner's perspective, which will be available in early 2025: “Legal Protection for Computer implemented Inventions – A practical guide for Software related Patent Applications".

In addition to her daily practice, she closely follows the current international developments of patent law in the field of legal protection for software.

Claudia is founder of the network women in ip and is member of the Management Board of the non-profit association ?women in IP“.

Claudia is going to speak at the German-language course on claims in times of AI and machine learning in Munich on 17 December 2024, together with Maria Sorea (Nokia) and Miriam Kiefer (Kather Augenstein).

The cover image was created with ChatGPT 4o.

The training and test sets image was shared under CC BY 3.0.