The Unified Patent Court (UPC) has issued an important decision in UPC_CFI_660/2024, a case between Centripetal Limited and Palo Alto Networks, Inc. The decision concerned European Patent EP 3 652 914 — a patent directed to AI driven methods for analysis workflows to detect cyber threats and malicious activities.
Whilst the technology is based on cybersecurity the decision is relevant to any organisations having, or developing, a portfolio with AI‑enabled inventions. The decision offers valuable insight into how the UPC is approaching inventions that rely heavily on machine learning processes such as computational processing, pattern recognition, and automated decision‑making.
Background to the case
Centripetal Limited brought an infringement action before the Mannheim Local Division of the UPC, asserting both direct and indirect infringement of EP 3 652 914 — a patent covering automated analysis of event logs, where algorithms determine the “reportability likelihood” of cyber events before sorting and transmitting them for further assessment. The defendant, Palo Alto Networks, counterclaimed for revocation, resulting in a dual dispute over both infringement and validity within the UPC framework.
The opt‑out previously registered for the patent was withdrawn in November 2024, bringing the dispute fully within UPC jurisdiction and illustrating the strategic consequences of opt‑out decisions in fast‑moving technology sectors.
Consistency with EPO decision making
A welcome point of the decision is that it confirmed the consistency between the UPC’s approach and the European Patent Office (EPO)’s established approach to machine learning based inventions, and software inventions in general. As the EPO’s approach is settled, and well understood, by following the EPO’s approach the UPC will offer certainty to patent proprietors regarding their portfolio.
The decision followed the EPO’s approach of requiring that computational or algorithmic inventions contribute to solving a technical problem using technical means.
The UPC’s analysis of the claims — focusing on the technical effect achieved in cyber‑analysis workflows — aligns with this principle.
Reassuringly, as is consistent with the EPO’s approach, the UPC did not treat algorithmic or AI‑type features as inherently patentable nor in a manner different to any other software-based invention. Instead, it evaluated the claims based on their functional technical contribution.
This alignment signals that companies developing AI‑related technologies can expect greater legal predictability between the UPC and EPO — a vital consideration when deciding whether to litigate before the UPC or national courts.
Key takeaways
- UPC and EPO are showing strong alignment on the treatment of AI‑related and algorithm‑driven inventions — meaning European patenting and enforcement strategies can be developed with improved confidence.
- Opt‑out strategy needs careful management, particularly for companies in fast‑moving fields such as AI, cybersecurity, fintech, and data analytics.
- The UPC is positioning itself as a major forum for AI‑related disputes, accelerating Europe’s role in shaping IP standards for AI technologies.
- Now is the time to review AI‑related portfolios, ensuring both EPO prosecution strategy and UPC litigation strategy are aligned.
Diego Black
AI group
This publication is a general summary of the law. It should not replace legal advice tailored to your specific circumstances.
© Withers & Rogers LLP January 2026
