There must be at least one named human inventor who makes a “significant contribution” to the invention in order to secure patent protection for an AI-assisted invention. So says the US Patent and Trademark office (USPTO) in its new guidance for filing patent applications for inventions where AI has played a role. 

Inventorship guidance for AI-Assisted Inventions, which came into force today (13 February), provides instructions to determine whether the human contribution to an innovation is significant enough to qualify for a patent when AI also contributed.

Whether AI can be named as an inventor on a patent application is a developing area of the law. The USPTO previously denied naming an AI as a sole patent inventor in the Stephen Thaler patent application, which listed the AI device DABUS as a sole inventor; this was in line with the findings of courts and IP offices in other jurisdictions like the UK, Europe and China.

Kathi Vidal, under secretary of commerce for the USPTO, said in a blog that the new guidance, crafted with the aid of comments from public consultation, “embraces” the use of AI in innovation and provides that AI-assisted inventions are “not categorically unpatentable”. To secure patent protection, however, there must be at least one named human inventor who meets the “significant contribution” requirement.

She continued that instead of considering whether or not the contributions of the AI system to an invention would rise to the same level of inventorship if those contributions were made by a human, the key question this guidance helps address is whether the “human named on a patent made a significant enough contribution to be named as an inventor”.

The guidance stems from an executive order from President Joe Biden last October which called on the office to issue guidance on AI and inventorship.

The guidance says that by focusing on the human contribution its approach “supports the USPTO’s goal of helping to ensure our patent system strikes the right balance between protecting and incentivising AI-assisted inventions and not hindering future human innovation by locking up innovation created without human ingenuity”.

Additionally, the guidance sets out specific examples of hypothetical situations with advice on how it would apply to those situations to further assist examiners and applicants in their understanding. These include developing a therapeutic compound for the treatment of cancer and a transaxle for a remote control car.

The guidance warns that determining whether a “natural person’s” contribution in AI-assisted inventions is significant may be difficult to ascertain, and “there is no bright-line test”.

It also points out that the guidance applies not only to utility patents, but to design and plant patent applications.

Benjamin Hsing, an IP partner at Washington DC-headquartered Venable, said on LinkedIn that the USPTO is not changing or modifying its duty of disclosure or duty of reasonable inquiry. It is, however, reminding applicants and practitioners that information regarding inventorship may be deemed material to patentability, and that for AI-assisted inventions, such “information could include evidence that demonstrates a named inventor did not significantly contribute to the invention because the person’s purported contribution(s) was made by an AI system”.

The guidance further reminds practitioners to inquire about “whether and how AI is being used in the invention creation process”, he added.