The integration of artificial intelligence (AI) into our business and personal lives is something that has been transformative over the past decade. As new technologies emerge and quickly evolve, so does the growing demand to protect them. AI technology is reaching a point where it can create solutions that its creators did not program it to do. Owners of such technologies are beginning to seek protection of these AI-created solutions and inventions through ordinary intellectual property channels.
However, a series of UK Intellectual Property Office (UKIPO), EU Patent Office (EUPO) and US Patent and Trade Mark Office (USPTO) decisions have refused patent applications for inventions which are claimed to have been created by AI. Their decisions tell us that a change in legislation is required if patent protection of an AI invention is something that is to be considered seriously in Europe and the US.
DABUS – AI Inventor
Historically, machine-learning systems are described as tools which are trained to perform certain tasks. Any invention generated as a result of these tasks would not usually be considered an invention of the machine as the outcome is known in advance.
DABUS is a patented AI system created by Dr Stephen Thaler. Known as a ‘Creativity Machine’, DABUS defies the historical understanding of what a machine-learning system is. Dr Thaler claims that DABUS has conceived inventions independently and autonomously through use of its neural networks, and without human intervention. The machine created a design for a food container and a device which flashes warning lights in a fractal-based sequence to provoke attention. Dr Thaler applied to patent the inventions in the US, EU, and UK, listing DABUS as the sole inventor.
Could DABUS be considered an inventor by the patent authorities?
The UKIPO and USPTO stated that they did not dispute the fact that DABUS created the inventions, whereas the EPO said it was a matter for national offices to decide. In each case, the reason for the application being rejected was the applicant’s failure to list a human inventor on the paperwork. The substantive issues, such as whether the invention is novel and inventive enough, were not assessed by any of the offices.
The applicant requested that the requirement to list a human inventor be broadened to incorporate an AI system or be waived entirely. This was rejected by the offices. The USPTO argued that this would ‘contradict the plain reading of the patent statutes that refer to persons and individuals.’
What does this mean for the patent system?
These decisions mean that under current legislation there is no possibility that an AI machine can be listed as a sole inventor for a patent in Europe and the US. The UKIPO recognised the applicant’s concerns, calling for further debate on this issue and potential changes to patent legislation. But this is only the beginning of what may be a long and complex re-organisation of the patent system.
There is now the worry that innovations will be withheld from the public simply because patent protection is not available. Legislators will be faced with the difficult task of incentivising inventors whilst balancing the rights of AI systems and the creators of AI.