Patent Criteria Family
AI and Machine Learning Patentability
Addresses specialized eligibility and patentability issues for inventions involving computational models, algorithms, or mathematically framed techniques. Emphasizes technical character and technical contribution.
Overview
AI-related patent review often turns on whether the invention produces a genuine technical contribution rather than merely manipulating data or reciting a mathematical method. Both the EPO and UK courts have developed specialized two-hurdle frameworks for assessing patentability of AI and machine learning inventions.
Topic 6.1: AI and Machine Learning Inventions at the EPO
The EPO and UKIPO apply a two-hurdle test when assessing the patentability of AI inventions. The first hurdle asks whether the mathematical method contributes technical character; the second covers all other patentability requirements, including inventive step under COMVIK, sufficiency under EPC Art. 83, and the Aerotel four-step test for technical contribution.
The EPO defines AI and machine learning as "computational models and algorithms for classification, clustering, regression and dimensionality reduction, such as neural networks, genetic algorithms, support vector machines, k-means, kernel regression and discriminant analysis." Guidelines for Examination in the European Patent Office (April 2025), Part G, Chapter II, 3.3.
📥 AI and Machine Learning Patentability — PDF
Licensed under Creative Commons Attribution-ShareAlike 4.0 International (2026 by QuickSilver IP)
Why This Family Matters
AI-related patent review often turns on whether the invention produces a genuine technical contribution rather than merely manipulating data or reciting a mathematical method.
Have Questions About Your IP?
These resources support general understanding of IP strategy. For guidance specific to your situation, start a conversation.
Get in Touch