Changes to the European Patent Office Guidelines for Assessing the Patentability of Computer Implemented Inventions following the Enlarged Board of Appeal Decision in G1/19

11/05/2022

The European Patent Office (EPO) Guidelines relating to the patentability of Computer Implemented Inventions (CIIs) have been significantly amended following the publication of the seminal Enlarged Board of Appeal decision G1/19 on the patentability of computer simulations, reported here in 2021. The update places the T641/00 “COMVIK” approach at the centre of the test for patentability of CIIs, and includes a new discussion of its application to Artificial Intelligence inventions.

T641/00 “COMVIK”

The guidelines have been sharpened up to reflect the recent focus on the Technical Board of Appeal decision T641/00 “COMVIK”. COMVIK is a well established decision from the “noughties” – the early 2000s – and has always informed the approach to software patentability questions, along with other decisions such as T258/03 “HITACHI”, and T172/03 “RICOH”.

The COMVIK approach is used to assess the patentability of inventions that are implemented in software and which may otherwise fall foul of the Article 52 EPC exclusion of “software as such” from the realm of patentable subject matter. It teaches us that patentable inventions are expected to solve “technical” problems, and that they must be defined in terms of “technical” features. Features that are not technical, such as features that relate to commercial or administrative considerations, do not contribute to the assessment of inventive step.

COMVIK defines two hurdles when assessing patentability, each hurdle being a necessary condition that the claim must meet. The first hurdle is relatively easy to overcome and requires that the definition of the claimed invention at least refer to technical means, like computer hardware or computer software. This avoids the claim being directed purely to abstract, non-industrially or non-technically applied concepts such as “algorithms”, “methods of doing business” and so on. The second hurdle is harder to clear, and is a special formulation of the EPO’s inventive step test. This requires that the technical features of the claim be identified, and an assessment be made as to whether the technical features (in the context of the claim as whole) solve a technical problem.

As a result of the COMVIK approach an invention like “a take-away delivery method using blockchain technology” would be patentable only if the defined technical features in the claim (defining blockchain technology) were new and solved a technical problem, not because blockchain technology had been applied to take-away food delivery. Similarly, an invention like “the generation of personalised advertisements for cruise-ship vacations using artificial intelligence” would only be patentable if the artificial intelligence itself was new and solved a technical problem, not because of anything to do with the problems faced by tour operators.

In any example of computer implemented invention patentability, the details of the claimed invention, and their relevance to the closest prior art are critical. As well as stronger COMVIK references, the updated guidelines include further examples of CIIs that are considered patentable. These all build on the theme developed in the G1/19 decision, that the claimed invention must (whether explicitly or implicitly stated) intrinsically address a technical problem. A number of examples are considered below.

Mathematical methods generally

The guidelines emphasise that “mathematical methods” that are not applied to control of a real world process may still be patentable if they are “motivated by the technical considerations of the internal functioning of the computer”. The EPO have in mind cases where “the mathematical method is designed to exploit particular technical properties of the technical system on which it is implemented to bring about a technical effect such as efficient use of computer storage capacity or network bandwidth”, as well as “assigning the execution of data-intensive training steps of a machine-learning algorithm to a graphical processing unit (GPU) and preparatory steps to a standard central processing unit (CPU) to take advantage of the parallel architecture of the computing platform”.

Of course such examples are already well known in a general sense, so any patentable claim defining the invention would necessarily need to define the specific implementation details of the mathematical method and how they are significant.

This change reflects a de-emphasising of the teaching in the Technical Board of Appeal T1227/05 “INFINEON” decision which had stated that a simulation of a noise affected circuit had been patentable because the noise model formed a technical feature limiting the claim to “an adequately defined class of technical items”. The point of patentability in INFINEON can now perhaps be better understood as relating to the efficient use of computer resources. Further examples of an improvement in computational efficiency qualifying as a technical effect, are provided in Part G, II, 3.6 of the guidelines.

Metrology and Sensing

The detection of physical properties using a series of sensors has often been regarded as patentable, providing the data captured by the sensors represents the physical state of an entity, and the entity and data can be regarded a part of the solution to a technical problem (it is not necessary for the claim to specify that the data actually be applied to the problem). The guidelines have been updated in this regard to state that:

Defining the nature of the data input to a mathematical method does not necessarily imply that the mathematical method contributes to the technical character of the invention”. However, “if steps of a mathematical method are used to derive or predict the physical state of an existing real object from measurements of physical properties, as in the case of indirect measurements, those steps make a technical contribution regardless of what use is made of the results”.

The guidelines therefore now emphasise that determining the physical state of something, using measured physical properties makes a technical contribution, though one would that would still need to be assessed in the context of the closest prior art.

Simulations specifically

The G1/19 decision made it clear that where the claimed invention was limited to a simulation, it was only necessary for patentability for the simulation to have technical character and solve a technical problem under the COMVIK approach. As such, questions of whether the simulation had a direct link to a physical reality, or was purely numerical, or was restricted to a well defined class of technical items were either inappropriate or secondary. The discussion in the Guidelines about the patentability of simulations has therefore been rewritten to focus on whether the simulation makes a technical contribution using the themes developed in G1/19. These include whether the claimed invention has “a potential technical effect” once it has been run on a suitable computer, or whether the invention “is specifically adapted for a technical use”.

Video Games

Interestingly the question of video games arose in G1/19 and a further example of when a computer implemented invention in the form of a video game may be patentable is provided.

“In contrast to effects that reside within the virtual game world or are otherwise inherent to the model already, a specific implementation of a simulation, if adapted to the internal functioning of a computer system, produces a technical effect. For instance, merely predicting the virtual trajectory of a billiard ball shot by the player, even if highly accurate, fails to solve a technical problem beyond its implementation. In contrast, adjusting the step sizes used in the distributed simulation of bullets fired in a multi-player online game based on current network latencies produces a technical effect”.

The pedestrian simulation considered in G1/19 had application to the safety aspects of the design of a public place like a stadium or railway station. However, patentability faltered because it did not necessarily solve the problem of safety, and could equally well have been used for modelling a crowd of people in a video game without reference to the resource availability or internal function of the computer. The Enlarged Board used a “billiards” simulation as an example of a non-patentable computer simulation modelling a real world situation. Here, the EPO appears to regard the control of “bullets” in a multi-player online video game as a computer resource intensive feature, and control of such a feature as potentially patentable. Again, patentability would be likely to turn on the details of the case.

Additive Manufacturing and 3D Printing

Another interesting example is the mention of “digital data” and its use to control devices in additive manufacturing or “3D printing” type applications. In the example, “since the data comprises both a digital description of the (physical) product of claim 1 and associated operating instructions adapted to control an AM device, it is intended to be used to control an AM device to fabricate the product”.  

Thus, a “technical use of the data is implied across substantially the whole scope of the claim” and the claimed invention is understood to make a technical contribution and be patentable. This would not be the case however if the claimed digital data did not imply its use, in other words merely provided a digital description or 3D model of a product not adapted to a manufacturing process, that could merely be used to visualise the produce in a CAD software tool.

Guideline Examples for the COMVIK approach

The present guidelines Part G, VII, Section 5.4 already include a number of examples showing how the COMVIK approach can be applied to different types of invention. Presently, these include (1) “a method of facilitating shopping on a mobile device”, (2) “a computer-implemented method for brokering offers and demands in the field of transporting freight”, and (3) “Transmission of a broadcast media channel to a remote client over a data connection”. A fourth example has now been added relating to environmental control systems: “a computer-implemented method of determining areas in which there is an increased risk of condensation for a surface in a building”, and a fifth example relating to the use of Artificial Intelligence in a thermal spray coating process. This is discussed below.

The decision referred to above, T1227/05 INFINEON, “a computer-implemented method for the numerical simulation of the performance of an electronic circuit subject to 1/f noise” provided a fourth example but has now been deleted.

New Guideline Example for the COMVIK approach: “Artificial Intelligence”

A new example has also been added to the COMVIK approach examples relating specifically to an application of artificial intelligence in a thermal spray coating process, in which if an actual value of the process deviates from a target values, the following step is performed:

“adjusting process parameters for the thermal spray coating process automatically by a controller on the basis of a neural network, said controller being a neuro-fuzzy controller which combines a neural-network and fuzzy logic rules and thereby maps statistical relationships between input variables and output variables of the neuro-fuzzy controller

Although applied to a technical process, the claim is found non-inventive over the prior art in this example, as:

“claim 1 does not contain any information about the coating properties to be achieved. The input and output variables of the neuro-fuzzy controller, how the controller is trained or how the output is used in the regulation of the process parameters are not defined. No features of the neuro-fuzzy controller are linked to any technical properties of the spray coating. The neuro-fuzzy controller is therefore not adapted for the specific application of thermal spray coating.

As ever, in order to imbue an Artificial Intelligence based invention with a technical character, it is necessary to define the technical nature of the relevant input and outputs. This makes it capable of forming a solution to a technical problem, and allows it to be distinguished from other (non-patentable) computer implemented mathematical methods.

Conclusions

The new guidelines are not radically different from earlier iterations, and the new additions reflect the recent thinking in G1/19, and its focus on the COMVIK approach as the primary means in all situations to resolve questions of software patentability and patent eligibility. The new examples do however provide interesting insights into how existing European patent law in this area is brushing up against new areas of technology, and how the European Patent Office analysis is evolving over time.

The take home message for applicants is the same. In order to secure patent protection at the EPO, it is necessary to consider how the claimed software or computer implementation addresses a technical problem : how does the claimed software makes the computer or network running the software operate better, or how does it reflect the existing realities of the hardware implementation and achieve a technical result within limited technical constraints? It is also necessary to make sure that the application and claim contain sufficient detail of the features that will be relied upon to make these points to the EPO during examination.

This is an area where we have considerable expertise and where we love to help. If you have any queries in this regard, let us know.

The marked up new guidelines are available here at the EPO website.

This article is for general information only. Its content is not a statement of the law on any subject and does not constitute advice. Please contact Reddie & Grose LLP for advice before taking any action in reliance on it.