As Artificial intelligence (AI) continues to evolve, its intersection with Intellectual Property (IP) law has become a crucial consideration for innovators.
The UK Intellectual Property Office (IPO) has a set of detailed guidelines to evaluate if AI inventions are indeed patentable. Even with this guideline, it still provides a new, unique set of challenges and opportunities that inventors need to be aware of as they start to implement and develop with AI. As legal advisors in the emerging technology space we often have to advise and draft documents predicting what regulations and the landscape might look like in the future. It is one area that we continue to assess on a regular basis.
In the first blog of these two articles I covered AI and Patent Law and the impact of AI in IP Law. In this second instalment, I cover strategies to protect AI innovations, future trends of AI and IP law, AI driven IP management and the international merger of AI patent IP laws.
In this first blog of two articles, I will cover AI and Patent Law and the impact of AI in IP Law. The second instalment will cover strategies to protect AI innovations, future trends of AI and IP law, AI-driven IP management and the international merger of AI patent IP laws.
AI and Patent Law
The UK IPO has set that AI is patentable and across a large area of fields. The patentability is largely under the regulations of the primary act section 1 (2) of the Patent Act of 1977.
Key Considerations for Patentability
- Technical Contribution: AI inventions need to demonstrate a technical contribution beyond that of a mere computer program or mathematical method. This means the innovation must solve a technical problem using technical means. For instance, an AI system that improves the efficiency of a gas supply monitoring system by detecting faults based on sensor data is considered to make a technical contribution and is patentable.
- Excluded Subject Matters: AI applications that are deemed to be methods of doing business, mathematics, or mental; are not patentable. For instance, an AI system for automatic financial trading would likely not be patented as it pertains to a method of doing business.
- Novelty and non-obviousness: For any patent, inventions must be novel and non-obvious. The invention should be sufficiently disclosed as to enable a person in the art to perform it. As defined in section 2 of the Patents Act of 1977, an invention is considered new or novel if it does not form part of the start of the art, meaning it must be a new invention. Secondly, it must be non-obvious. An invention is considered to involve an inventive step if it is not obvious to a person skilled in the art. This means that the invention must not be an obvious variation of what is already known.
Impact of AI in IP Law
AI has continued to impact various fields and is now integrating into all trades, businesses and sectors, and the related challenge of IP law is no exception. The IPO has set clear guidelines to accentuate the need for clear and concise distinctions between technical innovations and mere applications of AI methods being used within a business sector.
Scenarios and Their Implications
The IPO’s guidelines they have set out provide scenarios to illustrate the applications of these principles.
- Allowed AI Inventions: Systems that use neural networks to process and classify data for specific technical purposes, such as detecting cavitation in pump systems or controlling fuel injectors in engines, are not excluded because they solve technical problems using AI.
An example of this would be a smartwatch. If a tech firm wanted to patent a watch that detects types of motions to determine and identify what activity the user is doing, it would most likely not be excluded. The system to do so would require real-world data from the watch’s gyroscope, magnetometer, and accelerometer. Since the contribution is not just a program and classifies measured system data, it is concisely technical, making it not excluded under section 1(2).
- Not Allowed AI Inventions: AI systems for patient data analysis for medical need assessments are excluded because they are considered methods of doing business or administrative management tasks.
One example would be the use of an AI-trained classifier to identify and move junk email. This new method uses the users’ input of what they put into junk mail to learn what needs to be moved there automatically. The product’s contribution is a method of classifying emails based on text structure and content. This contribution, however, is merely a program for a computer and has no technical effect besides the running of a program. Meaning, under section 1(2), the invention is excluded.
This article will conclude in part two, available on 16th August 2024.
