Patents usually protect inventions that come from a human mind. What happens when the inventor is not human? If a computer designs something new, should it be named as an inventor?
Ryan Abbott, a law professor at the University of Surrey and a former USPTO staffer, argues that computers should be able to count as inventors. In a recent paper he calls for a basic shift: treat creative machines as inventors so the system rewards more inventive software and, in turn, more useful innovation.
Rights for people and machines
In a Singularity Hub interview, Abbott notes that today’s patent rules do not really admit non-human inventors, even when the output looks patentable. Peter Rejcek points to cases like the Oral-B CrossAction toothbrush design, plus some compositions and recipes, as examples that sit in this gray zone.
Abbott puts it this way: creative computers may force us to revisit what inventorship means, and maybe the patent system as a whole. If you have something that meets the bar, getting a patent is not the hard part. Naming the inventor is.
If the criteria are met, a patent should follow
Under current rules you still need novelty, non-obviousness, and utility. Abbott’s view is simple: if a computer or AI system produced something that clears those bars, the patent is earned.
One tidy fix is to grant the patent to the people who built the AI. That breaks down when several teams and several models collaborate on one artifact. Who owns the inventorship then? The law has not settled it, and Abbott says it will need an answer soon.
As AI capability climbs, he wants foresight, not denial: “AI will reshape labor markets and social life. People need to think this through and find ways for society to share the upside.”
