31 October 2018

Has patenting AI and computer-implemented inventions just got easier in Europe?

The European Patent Office (EPO) has just published the new edition of their Guidelines for Examination that will come into effect from 1st November 2018.

The Guidelines for Examination give instructions on the practice and procedure to be followed in the various aspects of the examination of European applications and patents in accordance with the European Patent Convention and its Implementing Regulations and is supposed to be followed by all European Examiners. Any changes to these Guidelines can therefore impact the action and conduct of Examiners in Europe. Key changes are outlined below.

Guideline changes

Alongside re-written sections on unity and a move towards more definitive language e,g. “should” changing to “is”, “must” and “needs to”, the Guidelines provide clarity on ongoing the question: what is technical with regards to computer-implemented inventions?

Perhaps most relevant to those working with AI and computer-implemented inventions is the creation of entirely new sections directed at, for example:

  • cases where the invention is realised in a distributed computing environment;
  • mathematical methods;
  • artificial intelligence and machine learning;
  • claims directed to methods of simulation, design or modelling; and
  • programs for computers.

AI and Machine Learning

The entirely new section on AI and machine learning explains how AI and machine learning find applications in various fields of technology and includes patentable examples such as the use of a neural network for identifying irregular heartbeats. While the new section does list some exceptions to what might be considered technical (for example the classification of text documents solely in respect of their textual content), encouragingly it does indicate that “where a classification method serves a technical purpose, the steps of generating the training set and training the classifier may also contribute to the technical character of the invention, if they support achieving that technical purpose”. This suggests that claims directed to methods of training AI, and datasets for use in such training, are patentable. This may have important implications for those working, for example, in the field of drug discovery.

The addition of the new section on AI and machine learning appears to follow in the wake of the EPO’s first conference on patenting AI nheld in May 2018 where the challenges and opportunities of patenting AI were discussed. The amendments to the guidelines to include this entirely new section appear to be indicative of a more favourable approach to AI being adopted by the EPO, in line with the remarks of Grant Philpott, the EPO’s chief operating officer for ICT, when he highlighted that the patent system needed to work very hard to ensure
it remains an opportunity for innovation in AI.

Mathematical Methods

In particular, the new section on mathematical methods provides a number of helpful examples of technical purposes which may be served by a mathematical method to help support the presence of an inventive step. These include:

  • separation of sources in speech signals; speech recognition, e.g. mapping a speech input to a text output;
  • encoding data for reliable and/or efficient transmission or storage (and corresponding decoding), e.g. error-correction coding of data for transmission over a noisy channel, compression of audio, image, video or sensor data;
  • encrypting/decrypting or signing electronic communications; generating keys in an RSA cryptographic system;
  • optimising load distribution in a computer network;
  • determining the energy expenditure of a subject by processing data obtained from physiological sensors; deriving the body temperature of a subject from data obtained from an ear temperature detector;
  • providing a genotype estimate based on an analysis of DNA samples, as well as providing a confidence interval for this estimate so as to quantify its reliability; and
  • providing a medical diagnosis by an automated system processing physiological measurements.

Similarities to the US

In some ways the inclusion of such examples in the EPO Guidelines is similar to the practice of the US Patent and Trademark Office (USPTO) issuing examples for evaluating claims for patent subject matter eligibility under 35 U.S.C. 101.  Many US patent practitioners working in this technical sphere already work to draft and prosecute their applications so that, at least in the minds of the Examiners, their patent applications will fall within one of these safe harbours. With the issuance of these new examples in the EPO Guidelines for examination, perhaps we will see the same being done by European patent attorneys.

What this means for Innovators

Whether these changes to the EPO Guidelines will make patenting AI and computer-implemented inventions as such easier in Europe is difficult to say yet.  At the very least, the changes should make prosecution of patent applications in these areas more predictable. Therefore it’s best to seek advice from experienced European patent attorneys at an early stage, so they can help draft and prosecute patent applications in these technical areas in a manner that will navigate these safe harbours and improve the prospects of obtaining patent protection in Europe.

This article was first published in Intellectual Property Magazine. 

Additional commentary from Andrew White is also featured in Global Data Review.

To discuss protecting the IP in your AI or other technical innovation, please contact our IT&E team today.