Papula-Nevinpat patent attorney Tomi Konkonen has been working with information technology and electronics patenting in Europe for more than 20 years. He explains how to overcome some of the challenges associated with patenting software inventions in the era of artificial intelligence and the Internet of Things.
“You sometimes hear people saying that it’s not possible to patent software. But it is!” says Konkonen, speaking as part of a Papula-Nevinpat webcast. “In Europe, you can do so if there is what we call ‘an inventive technical contribution’ to the prior art.”
“The European Patent Board has a very systematic approach that has been generated by long-established appeal board decisions,” he says. “Any software invention in Europe must overcome two hurdles in order to be patentable.”
The first such hurdle concerns so-called ‘excluded subject matter’, whereby harmonized laws in Europe have barred certain things from being patentable. Although software falls into this category, the hurdle can easily be overcome by including just a single technical feature in the patent claim that shifts it out of pure software territory and into the domain of traditional engineering too.
The second hurdle is more difficult to overcome. It concerns the evaluation of the software’s so-called ‘inventive step’, whereby only technical features that contribute to the prior art are considered. All non-technical details are consciously omitted as part of this stage of the patent evaluation process.
“The inventive-step approach means that your work must be something completely new in the engineering world, even though it is software and has abstract ideas behind it,” says Konkonen.
Navigating a thin line
So what does the European Patent Office (EPO) consider to be technical versus non-technical when it comes to artificial intelligence and the Internet of Things? Konkonen says the answer is not straightforward.
“Originally, our patent systems were designed for nuts and bolts inventions – not software – so the framework doesn’t work perfectly in the modern world,” he says.
Given the thin line between what is patentable or not in software, the best way to understand the environment is to look at practical cases. For example, a neural network that uses unique algorithms to identify irregular heartbeats would be patent eligible, as it makes a new technical contribution to the field. But computational models and algorithms on their own – without a technical framework within which to operate – would not be patent eligible.
“IoT inventions typically meet the requirements of the EPO’s ‘excluded subject matter’ framework, but the ‘inventive step’ clause often causes issues,” says Konkonen.
“For instance, a connection between a toothbrush and a home network would not be patentable. But if the toothbrush were to intelligently know which tooth needs more brushing – and if this could be illustrated through an app interface – then that would indeed be possible to file a patent for.”
Be specific and be fast
Konkonen explains the importance of patent claims having detailed information on technical elements, applications and the problem being solved. Claims should also include a single infringing actuator. For example, instead of describing an invention as “a communication system between an elevator and a mobile phone”, the language should rather be structured in such a way as to elicit a single actuator that it would be easier to identify an infringement upon, namely: “an elevator comprising a communication system that uses a mobile phone.”
“With inter-connected inventions such as we see in the IoT world, it’s very important to specify where the so-called ‘point of novelty’ lies,” says Konkonen. “If it’s in an IoT device itself, then it’s clear that you cannot extend your patent to cover the server – you can only hope to patent the device.”
Konkonen has concrete advice for IoT start-ups, many of which have limited resources for filing patent claims and identifying infringements.
“One key criteria these days is speed. Try to get your patents filed fast, as the lifespan of software-based products is very limited,” says Konkonen.
It’s impossible to monitor globally for patent infringement, so the most effective way to do so is to keep up to date with the work of your clients and your closest competitors,” he says. “Monitoring in this way – as part of your daily business practises – is the most efficient way to identify patent violations.”