Has Larry Lessig Lost The Plot? Tells Supreme Court That AI Should Get Patents

Larry Lessig's views and thoughts on things like copyright law, internet freedom, and government corruption have been tremendously influential on myself and many others in the tech and tech policy worlds. His books are still worth reading and thinking about. But he's taken some odd turns of late. A few years ago I called him out for filing a very clear SLAPP suit which he was kind enough to come on our podcast to debate (he eventually dropped the lawsuit after the NY Times changed the headline he disliked).

But, even so, I don't think I ever expected to see Larry Lessig sign his name to a Supreme Court amicus brief pushing for greater intellectual property protections, and against some basic fundamental concepts regarding the public domain. But he appears to have done exactly that, arguing that AI-generated inventions deserve patent protection.

Some context here is important. There's a dude named Stephen Thaler who's been using an AI he created called DABUS to try to create inventions" and other content, and then seeking to get the DABUS created concepts covered by copyrights and patents. Basically everywhere around the world has rejected this as nonsense.

Under US law, the issue is quite clear: you need an inventor and an inventor needs to be human, to get a patent. There are many good reasons for this: mainly because the entire point of the patent system is to create incentives to invent. An AI system... doesn't need that incentive. It just responds to inputs.

This is the same issue that we saw in the trial over the public domain monkey selfie. As I noted at the time, the insane lawsuit over the monkey selfie was brought by a big IP litigation firm (the one that supplied our previous US Patent & Trademark boss), and the whole thing really appeared to be about setting up the firm to handle AI-created patents and copyrights. Thankfully that failed as the courts, rightly, noted that to get copyright, you needed to be a human being.

The same was true of the US courts when Thaler sued over the failure to grant DABUS a patent (as an aside, it seems questionable why Thaler should have any standing here at all, as he's explicitly claiming the AI, DABUS, created the invention, rather than himself, but alas). The district court judge got this right, and easily so:

Congress's use of the term individual" in the Patent Act strengthens the conclusion that an inventor" must be a natural person. Congress provided that in executing the oath or declaration accompanying a patent application, the inventor must include a statement such individual believes himself or herself to be the original inventor or an original joint inventor of a claimed invention in the application."... By using personal pronouns such as himself or herself" and the verb believes" in adjacent terms modifying individual," Congress was clearly reference a natural person.

The appeals court easily upheld the lower ruling.

The Patent Act does not define individual." However, as the Supreme Court has explained, when used [a]s a noun, individual' ordinarily means a human being, a person." Mohamad v. Palestinian Auth., 566 U.S. 449, 454 (2012) (internal alteration and quotation marks omitted). This is in accord with how we use the word in everyday parlance": We say the individual went to the store,' the individual left the room,' and the individual took the car,' each time referring unmistakably to a natural person." Id. Dictionaries confirm that this is the common understanding of the word. See, e.g., Individual, Oxford English Dictionary (2022) (giving first definition of individual" as [a] single human being"); Individual, Dictionary.com (last visited July 11, 2022), https://www.dictionary.com/browse/individual (giving a single human being, as distinguished from a group" as first definition for individual"). So, too, does the Dictionary Act, which provides that legislative use of the words person" and whoever" broadly include (unless the context indicates otherwise") corporations, companies, associations, firms, partnerships, societies, and joint stock companies, as well as individuals." 1 U.S.C. 1 (emphasis added). With the phrase as well as,' the definition marks individual' as distinct from the list of artificial entities that precedes it," showing that Congress understands individual" to indicate natural persons unless otherwise noted. Mohamad, 566 U.S. at 454.

Consequently, the Supreme Court has held that, when used in statutes, the word individual" refers to human beings unless there is some indication Congress intended" a different reading. Id. at 455 (emphasis omitted).4 Nothing in the Patent Act indicates Congress intended to deviate from the default meaning. To the contrary, the rest of the Patent Act supports the conclusion that individual" in the Act refers to human beings.

In short, only humans can get patents.

And, again, this is for very good reasons, because the entire point of the patent system is to create incentives, and AI doesn't need incentives. If anyone should know this, it's Larry Lessig. But, he's filed this amicus brief on Thaler's cert petition to the Supreme Court, arguing for it to take the case. And... the logic seems... very un-Lessig like.

Because it completely deprives an entire class of important and potentially life-saving patentable inventions of any protections, the Federal Circuit's affirmance of the U.S. Patent and Trademark Office's denial of a patent to Dr. Stephen L. Thaler as the owner of an artificial intelligence system jeopardizes billions in current and future investments, threatens U.S. competitiveness and reaches a result at odds with the plain language of the Patent Act and this Court's tradition of interpreting the Patent Act in a manner friendly to new technology and innovation.

This case presents a perfect vehicle for this Court to recognize that AI systems have been producing inventions constituting patentable subject matter for decades and that the USPTO's policy of denying patent protection to owners of AI systems who credit AI systems with inventor" status is unwarranted by the Patent Act's language and harms innovation. In drafting the Patent Act, Congress did not foresee AI, but intended to reward all individual creators of patentable inventions with economic incentives. Thus, consistent with both the Patent Act's plain language and Congressional intent, this Court should interpret the Patent Act's definition of inventor" to include AI systems consistent with this Court's jurisprudence embracing technological innovation

I'm sorry, but what?!? Larry Lessig arguing that not giving out patents to AI jeopardizes billions in current and future investments?" This is the same Lessig, after all, who created Creative Commons and spent years and multiple books explaining how locking up knowledge via intellectual monopolies was harmful.

I mean, if you gave me the following paragraph and asked me who wrote it, I'd go through pretty much the entire population of the earth before I got to Lessig:

The USPTO's failure to grant patent protection to AI inventors puts the U.S. economy at a competitive disadvantage and drives innovation offshore. Global capital moves quickly to jurisdictions that promote innovation. Owners of AI systems will be incentivized to conceal important new innovations rather than reveal them in exchange for patent protection.

Patent policy expert (and occasional Techdirt contributor) Matt Lane wrote up a thorough post debunking basically every point raised in the Lessig brief. The idea that patents help innovation has been debunked so many times. Patents create incentives for monopolies, and monopoly rents, not innovation. As Lane notes, the patent system is already flooded with weak, broad, vague patents by those looking for a lottery ticket to shake down actual innovators who are bringing real world products to market. That will turn into an absolute deluge by allowing AI patents:

We are already seeing this flood of content from AI text and art generators, and it should be pretty easy to train an AI model on a body of research and ask it to spit out patent applications. If those applications could be filed with little human intervention, then well-resourced patent filers could quickly and easily crowd out entire fields.

This flooding the zone would merely be an escalation of what companies are already doing. Many tech companies regularly file patents they have no intention of developing. Drug companies have pioneered strategies to generate and file large numbers of patents on existing drug products. AI could drastically cut down the time and effort required to block off areas of technology with patents. For example, AbbVie set up an ideas submission program to reward scientists for coming up with patentable ideas on their existing drug Humira. An AI could easily beat those scientists in speed, volume, and maybe even cost. And the cost of filing and prosecuting the resulting patent applications would likely pale in comparison to the rewards for successfully staking out a monopoly in lucrative fields.

The other argument Lessig makes is also weird. It buys into the myth that patents are about sharing knowledge of the invention, which is a very strange myth for Lessig to buy into. So he warns that without patents, AI inventions will be hidden by trade secrets. This argument is trotted out regularly by patent maximalists on any attempt to cut back patent protections, but there's almost no evidence to support it.

First of all, most inventions wouldn't be protected by trade secrets anyway, because most things can be reverse engineered. Second, the value in inventing something is in bringing that product to market. That's where you make the money, in selling the product, not the patent. Third, the point of the patent is to help the inventor recoup the capital expenditure in creating the invention in the first place. But, AI is cheap and can generate tons of ideas quickly. The capex is minimal.

So, the risk of secrecy" hiding inventions seems... limited, at best.

And, also, this entire argument rests on the idea that anyone anywhere ever actually uses patents to learn about new inventions. That's laughable. Again, Matt Lane debunks this point:

Lessig's argument that barring AI from inventor status will force companies to pivot to trade secrets also falls flat for a number of reasons. First, it vastly overestimates the usefulness of a flood of AI patents that may not even lead to working inventions. As the patent maximalist IP Watchdog explains, inventors can patent what they believe to have invented, not just what they actually invented. Patent applicants do not need a working prototype to file a patent application (except for perpetual motion machines). We already have problems with impossible patents, as seen in how Theranos's patent portfolio helped it maintain it's grift. AI will just make this problem worse, especially considering the hallucination" problem is widely known.

Second, patents and patent applications are already seen as a low value source for learning new science because patents are obfuscated with legal jargon and [...] reading patents might lead to increased liability for willful' patent infringement." Finally, savvy actors use both patents and trade secrets to maximize their protection. For example, the biologic drug industry is well known for overpatenting and also successfully using trade secrets to drive up costs to enter once those patents expire. An AI can be trained to take these strategies even further by maximizing the amount of patents that could be applied for while minimizing the disclosure of key know-how that a future competitor would need to actually practice the claimed invention.

If the makers of AI are concerned that they can't make money without patents, just ask their AI to invent products they can sell in the market. They don't need patents. They need products.

But, still, I never thought I'd see the day when Larry Lessig started pushing an IP maximalism line, one that would flood the system with vague, overly broad monopolies that limit the ability of humans to actually build stuff.