A US appeals court has ruled that an artificial intelligence system cannot be an inventor under the United States (US) patent law. The US Court of Appeals for the Federal Circuit said the Patent Act needs the inventor to be a natural person. The ruling rejected computer scientist Stephen Thaler’s plea for patents on two inventions created by his DABUS system.
Thaler said that DABUS, i.e., Device for the Autonomous Bootstrapping of Unified Sentience, is sentient and natural. His attorney Ryan Abbott said the decision undermines the purpose of the Patent Act and has tangible negative social consequences.
The US Patent and Trademark Office (PTO) refused to comment on the decision. Thaler has lost other bids for patents that state DABUS as their inventor in Australia and the European Union. A Virginia court and PTO, both rejected Thaler’s applications for DABUS patents because the system is not a human being.
Read More: US Copyright Office Sued Over Rejection To Grant Authorship To AI Model
Thaler challenged the Virginia court decision before the Federal Circuit, which deals with patent appeals. Abbott told the Federal Circuit that the ruling was at odds with the plain language and purpose of the Patent Act. He added that the act is meant to promote innovation and does not explicitly mention that an inventor must be a natural person.
However, in a unanimous three-judge panel, Circuit Judge Leonard Stark said there is no ambiguity that the Patent Act requires inventors to be natural persons, i.e., human beings.
Stark said that Thaler’s argument that awarding patents to artificial intelligence systems would encourage innovation was speculative. He also dismissed Thaler’s concerns that denying AI patents would diminish the purpose of patents outlined in the US Constitution.
There have been instances where AI has been recognised as a patent inventor. In the month of February, artificial intelligence-powered facial recognition system developing company Clearview Ai received a US patent for its revolutionary face recognition platform.
