Can AI machines be inventors or authors? The US says no—for now—but the AI juggernaut won't stop

Two recent cases in the US—both brought by the same plaintiff, an inventor, Stephen Thaler—have brought further clarity to patent and copyright law as it relates to creations by artificial intelligence programmes.

And those same cases—broadly accepted by lawyers to have been decided correctly under the current law—raise deep and important questions about whether AI programmes should be allowed to be deemed as the inventor of new technologies for patents, or the author of new works for copyrights.

In the first, Thaler created the DABUS (Device for Autonomous Bootstrapping of Unified Sentience) machine, which Thaler used to create an invention that Thaler then applied for a patent with the US Patent and Trademark Office (USPTO).

The USPTO denied the patent application because Thaler named DABUS as the inventor of the underlying technology.  The US Patent Act requires that an inventor be an “individual” who invented or discovered the subject matter of the invention.  Thaler appealed the USPTO’s decision and, ultimately, the US Court of Appeals for the Federal Circuit decided that the term “individual” refers to a natural person—a human being—and not to a machine.

Similarly, in Thaler’s copyright case, Thaler used the DABUS machine to create a piece of art.  Thaler submitted the artwork—“a pixelated pastoral scene of train tracks running under a moss-flecked bridge”—to the US Copyright Office for registration.

In August 2023, the US District Court for the District of Columbia, like the court in Thaler’s patent case, determined that no copyright could be registered with respect to the artwork because there was no human authorship to attach the work to.

Thus, the Copyright Office found, no valid copyright ever existed or could exist because the work was “generated absent human involvement[.]” [Thaler filed an  appeal on 11 October].

Most lawyers agree that the courts’ decisions accurately reflect the language, spirit, and intent of the Patent Act and the Copyright Act.  But, in this age of generative AI, should it?

DABUS hints at AI future

It’s widely understood that the purpose of the Patent Act and Copyright Act is to encourage and incentivise creation—by humans—to “promote the progress of science and useful arts” (cleaned up).

And, before there were widely available generative AI machines for use by humans, the only meaningful way to encourage and incentivise such creation was to reward the human beings who were directly responsible for creating the output.  (The Patent Act and Copyright Act incentivise and encourage such creation by awarding creators the exclusive use—a monopoly—of their creations for a limited period of time.)

Yet, Thaler’s creation of his DABUS machine, which is capable of creating new inventions, and new useful arts, shows precisely how human beings would be incentivised to create and use tools that help them invent new technologies and author new works.

As Thaler argued in his brief in the patent case, rejecting patents for inventions created by AI machines “curtails our patent system’s ability—and thwart’s Congress’s intent—to optimally stimulate innovation and technological progress.”

That is probably true. The new technologies and works potentially created by advanced AI machines may be wondrous, and potentially beyond the capability of human beings to create, or at least create now.  For example, Thaler’s DABUS machine invented “a food container with a fractal surface that helps with insulation and stacking” and “a flashing light for attracting attention in emergencies.”

Of course, those devices fall on the quotidian side, but the very fact of their usefulness and potential for improving human lives hints at a future filled with useful, novel, AI-generated technologies that are worthy of patenting.

US patent system favours GenAI platforms

One important component of patent laws in the US is that, in exchange for the inventor getting a limited-time monopoly over their patent, the inventor must disclose the technology with enough specificity to allow others to recreate the technology and build upon it, further enhancing innovation.

The Patent Act and Copyright Act aim to maximise scientific and artistic progress by trying to balance the incentives for innovation against the downside of restricting use by the broader public to new writings and discoveries.

In the case of AI-generated inventions and works, it is arguable that that balancing sharply favours incentivising the humans who invent generative AI machines, and humans who use generative AI machines to invent technologies and author works, to keep using AI machines to create new inventions.

One way to create that incentive is to allow the machines themselves to be inventors and authors, in which case the humans who create and own those machines can be more directly and clearly incentivised to create and use them.

There are, of course, countervailing concerns related to this proposition. Without the limitations of human beings, AI machines could potentially create, in essence, infinite inventions and infinite copyrightable works.

Such inventions and works could effectively “occupy the field” of creation, crowding out inventions and works made by human beings, and thus discouraging (and possibly making impossible) new patentable or copyrightable works by actual human beings.  But there are ways to limit such downsides in ways that will continue to encourage humans to create new inventions and art, while also encouraging humans to create AI machines that also create new inventions and art.

The US Congress is debating important issues relating to AI and its place in US law. Congress should carefully consider the purposes of the Patent Act and Copyright Act, and weigh whether allowing for AI machines to be inventors and authors further enhances the intent of those laws.

Mark Doerris a litigation and IP partner in the New York office of Greenspoon Marder. He can be contacted at:  mark.doerr@gmlaw.com