AI inventorship denial not surprising but is step towards change

AI inventorship

Reading time: five minutes and 25 seconds

In-house counsel from tech, automotive and pharmaceutical companies say it is business as usual after the EPO rejects AI inventor applications

The EPO’s rejection of two artificial intelligence inventor patent applications comes as no surprise to in-house IP lawyers, even though several acknowledge that this development could be the first step in longer-term changes to patent law.

Counsel from five companies working on AI solutions say the decision is in keeping with European Patent Convention (EPC) requirements that a human being must be listed as the inventor, and the fact that the technology is still years away from generating novel inventions without human input.

In an issued statement last month, the EPO justified its rejection by claiming that the two patent applications, submitted by the University of Surrey on behalf of an AI tool named DABUS (Device for the Autonomous Bootstrapping Of Unified Sentience), did not meet the necessary requirements for inventorship.

The chief patent counsel at a US tech company tells Patent Strategy that the decision will have little influence on a practical level on how his business operates because AI-generated ideas ultimately require a human inventor.

“Just like a hammer, chalkboard or an ordinary computer, AI is a tool that the inventor uses, and it doesn’t displace humans. At some point, a machine may get self-awareness, but that is decades away in my mind,” he says.

He adds that he is not surprised the EPO rejected the applications because the examiners are simply following the guidelines that are given to them: “If they say the inventor must be a human, you have no choice but to reject the application.

“I know there are some people who say that needs to change. But at whatever point it might need to change, I don’t think it can be changed by a patent examiner.”

The head of IP at a European pharmaceutical company agrees, and adds that the EPO was correct in denying the applications because an AI system has no legal personality.

“If something has no legal personality, it cannot own property like a patent. An AI cannot license a patent. For that, you need to be a person.

“Under the current legal framework, I do not think AI can be protected with a patent, but it is in the interest of patent offices to channel this into IP law,” he says.

For Gareth Jones, vice president of IP at BenevolentAI in the UK, allowing AI inventorship is important because companies like his rely on patents for licensing and depend on technologies to develop lifesaving medicines.

“The pharma industry works by investing huge sums of money in expensive and high-risk research with the expectation of a patent to protect exclusivity. A business like BenevolentAI, which uses technology for drug discovery, has to invest a lot to protect that technology. And if there is no ability to protect the output, that ruins the business model.”

He adds that inventions like the ones filed by the University of Surrey prove that machines are at the stage where they can be inventors, and that if machines are driving innovation, their inventions should be patentable.

He says: “How you change the law is up for debate. Whether you allow machines to be inventors or change the definition of what invention means so that humans still count as inventors are two avenues to explore.

“But people are confusing ownership with inventorship. Ultimately, ownership is much more important than inventorship because it is not the inventor who owns the patent but his or her employer.” 

Jones says that a future consequence of not allowing AI inventorship is that companies like his could rely more on trade secrets, which could result in slowing down innovation by frustrating the system that makes inventions publicly available.

Looking ahead

Ryan Abbott, professor of law and health sciences at the University of Surrey, who led the DABUS team, says it would have been a surprise if the EPO had accepted the applications without debate. He says this rejection is only the beginning in what he sees as a long road ahead for deciding the legal framework for AI-generated inventions.

“This is the first time a case of this nature has been brought to the attention of a patent office and it requires forward-thinking interpretation of laws not explicitly designed for these issues. We have anticipated that judicial intervention would be required and potentially legislative intervention too.

“The application was rejected by an EPO formalities officer and this is now being appealed to the Boards of Appeal. The appeal process can take years,” he says.

Even though the EPO rejected AI inventorship, there are signs that patent offices around the world are beginning to look more closely at how technology is changing the way companies innovate. Last year, the USPTO made a public request for comment on patenting AI inventions, and WIPO issued a draft paper on AI and IP last December.

James Dottavio, assistant director of technology transfer at the University of West Virginia, says that although he anticipated that the EPO would reject the DABUS patent applications, he is interested to see how the question of AI inventorship will evolve in the future.

“Everybody has been anticipating something like AI inventorship, both in Europe and in the US. There will need to be legislation if this is going to change. This decision will provoke some additional thinking and maybe some legislation. I don’t think the lack of protection will stop people using AI tools to help inventors,” he says.

The chief patent counsel at the tech company tells Patent Strategy he doesn’t believe denying applications to AI inventors is going to slow down innovation because machines are still reliant on human beings to operate. When machines become autonomous, he believes it will be time for legal changes to reflect the new realities.

He says that although patent offices are updating their guidelines and having public hearings about AI, he doesn’t know of many businesses that are asking for any significant changes in patent law.

“Right now we are seeing patent offices update their guidance or give public consultations to get a broader range of views and ensure that they are ready to cope with advances in AI. But I don’t see any of the patent offices getting loud calls from businesses to change the law so that no human is needed for innovation,” he says.

Whichever way the question of AI inventorship goes, it is clear that the recent rejection of AI patents is only the beginning of a very long battle to define rights, ownership, and the very need to incentivise innovation.

In the meantime, the decision has little practical impact on businesses – even those that are working closely on using AI to help come up with ideas.