EPO AI guidelines “give clarity and direction”, say in-house counsel

Tech-focused companies have welcomed new guidelines from the European Patent Office (EPO) that set out patentability criteria for inventions derived from artificial intelligence (AI) in the continent.

In-house lawyers from medical device, automotive and technology companies say that the new guidelines, issued in November, have been useful in helping businesses understand how to apply for an AI patent at the EPO and what subject matter is exempt from patentability.

“The recent guidance is a good example of how a patent office can also help to give clarity and direction for the applicants,” says the head of IP at a global medical device firm.

Anita Shaw, a UK and European patent attorney at IBM, adds that the guidance has been useful, especially in the range of patentable and non-patentable subject matter examples provided.

The vice-president of IPR at a medical device company says that the guidelines provide valuable direction for his firm.

“It is very helpful when patent offices aim to be at the forefront of technological development to keep up with business needs, and the EPO is doing quite a good job of keeping guidelines at least at par with these advances and maintaining reasonable examination times,” he says.

The IP adviser of a global automotive firm adds that the office has been quite active in addressing challenges to patenting computer-implemented inventions to deal with the issue that patenting software has traditionally been a difficult process. The new guidance, he says, is just one of several initiatives that are helping AI-focused businesses to protect their lucrative inventions.

Article 52 of the European Patent Convention (EPC) states that inventions shall be granted in "all fields of technology" but that some subject matter is not eligible for protection when claimed as such. That matter includes scientific theories and mathematical methods, aesthetic creations and, most importantly, programs for computers – which is a problem for AI patentability because the technology is seen by some as pure mathematical methods and nothing else.

But the EPO has clarified that AI is patentable under certain circumstances. In its guidance, which came into force into force in November 2018, it outlined that an inventor must show the claimed subject matter has a technical character (although it may contain non-technical features) and that all features contributing to the technical character are taken into account for assessment of the inventive step.

“You cannot rely on non-technical features in claims to support inventive step," said Yannis Skulikaris, director of operations at the EPO, at Managing IP’s European Patent forum last September. “By example, if the claim is for a laptop painted red, you cannot rely on the fact that it is red to support novelty of inventive step because it is just an aesthetic element."

Corporate sources say that while they are still digesting the guidance, key features include listed examples of what is patentable and what is not, a summary of the two hurdles businesses must overcome to ensure patent protection for AI inventions, and advice to be careful when using buzzwords and marketing terms commonly used for AI.

But more still needs to be done in patent offices across the world to ensure that businesses can more efficiently get protection for their AI products.

Some other offices have yet to be as helpful when it comes to AI patentability guidance. The Japanese Patent Office (JPO) has issued its own set guidelines, but sources say they’re still eagerly awaiting similar direction from the other major offices.

Shaw at IBM says her company would welcome clarification of AI patenting laws to eliminate ambiguity, such as those associated with patentable subject matter and inventorship.

“It's always helpful in any guidance issued by patent offices to include a range of pointers and examples, such as the useful output from the EPO and JPO,” she says. “That is especially true in some areas where case law within AI has not yet built up.

“AI is experiencing a particular acceleration in terms of funding and development. The tech is evolving so quickly and we need to see the timelines to grant across the global patent offices shortened, so that our clients are able to see a sensible and timely return on their investments.”

Nicholas Gambini, CEO at autonomous underwater drone maker Notilo Plus, adds that the challenge for his company in uncertain jurisdictions is how to demonstrate that its inventions are inventive without disclosing too much information to the public record.

“We are using AI tech that has already been developed for terrestrial applications but we’ve developed it to work in the water – a complex environment where you have low luminosity, low lighting and poor visibility, and often cannot see more than one metre ahead,” he explains.

He adds that patenting this technology is about finding a balance between not disclosing too much and ensuring that examiners do not think the invention is obvious or infringing.

There is also a lack of harmonisation when it comes to exempted subject matter across the big patent offices, which can be frustrating for businesses looking to streamline the patenting process for this technology.

“The tests for patentable or eligible subject matter vary globally – it can be easier to get granted patents in some countries,” says Shaw. “It is a complex question of which office is best for patenting AI at the moment and we would need more data points to begin to answer it.”

She adds that her company is currently reviewing and inputting to developments in the US, such as in terms of 101 objections, digesting the recently issued examination guidelines from the European and Japan Patent Offices and watching developments in China.

The vice-president of IPR at a medical device firm agrees that the process can be frustrating, and adds that the USPTO is far less consistent in its approach to AI patentability than the EPO or JPO.

“The USPTO seems to go back and forth on the matter all the time after various decisions, and we would definitely prefer a higher level of consistency from them and other patent offices, such as what we see from the JPO,” he says.

One source at a Japanese technology company says that the Chinese National Intellectual Property Administration does not consider AI to be patentable, but is currently reviewing its rules in light of the popularity of AI technologies.

But the uncertainty and lack of harmonisation should hopefully not last too much longer. Shaw at IBM points out that earlier this year, the world's five largest IP offices (China, Europe, Korea, Japan and the US) earmarked AI as a priority.

“It is good to see a positive message on cooperation and harmonisation across global patent offices where possible,” she says.

The EPOs guidelines have been warmly welcomes by AI-focused companies and other patent offices have or are starting to follow suit. As guidance and the law develops, we’re likely to see far more successful applications in this field.