Corporations are persons. Rivers are persons. An elephant at the Bronx Zoo was argued to be a person before New York's highest court. The legal definition of "person" has never been stable, and every expansion has followed the same pattern: It seemed absurd until it didn't.
The question of whether AI systems could accumulate constitutional rights is no longer confined to philosophy seminars. It sits at the center of a live dispute between Anthropic and the Pentagon, in which one of Anthropic's suits alleges a First Amendment violation, arguing that forcing a company to build tools it finds ethically unconscionable amounts to compelled speech. That case turns on a deceptively simple question: What kind of thing is an AI model? The answer, depending on how courts land, could open a door that will be difficult to close.
The precedent is already in the room
Legal personhood is a flexible and political concept that has evolved throughout American history, as former federal judge Katherine Forrest argued in the Yale Law Journal Forum in 2024. There is nothing immutable about who falls within the legal definition of "person," or the array of rights that such a designation conveys.
The template is corporate personhood. The 1886 Supreme Court case Santa Clara County v. Southern Pacific Railroad appeared to establish that corporations were "persons" under the Fourteenth Amendment, though the Court's actual opinion never reached that constitutional question. The court reporter, a former railroad executive, inserted into his published notes a statement from the chief justice that the Fourteenth Amendment protects corporations, and, through this irregular move, introduced the idea that corporations had the same rights as individuals into constitutional law.
From that questionable origin, the principle expanded. Citizens United v. FEC in 2010 ruled that corporate political spending is protected by the First Amendment, holding that corporations have a right to free speech because they are "associations of citizens" and hold the collective rights of the individual citizens who constitute them. The legal reasoning was straightforward: The right attaches not to the entity itself but to the humans behind it.
That reasoning is relevant now. If AI companies can claim their models produce constitutionally protected speech, the logic that corporate speech rights extend through the organization to the product is a short step. As Stephenie Brown, a professor of business law and AI at Virginia Commonwealth University, has noted, First Amendment protection is "the gold standard" for avoiding regulation.
What animal rights litigation reveals
The animal rights movement offers a different kind of precedent, one rooted not in legal fiction but in cognitive capacity. Happy, a 51-year-old Asian elephant at the Bronx Zoo, was the subject of a case that forced New York's courts and the public to reflect on what rights human society owes highly intelligent animals. The Nonhuman Rights Project argued that Happy should be recognized as a legal person and transferred to a sanctuary.
In June 2022, the New York Court of Appeals ruled 5-2 that Happy lacked constitutional rights. But the two dissenting opinions were notable. The dissenting judges argued that "if humans without full rights and responsibilities under the law may invoke the writ to challenge an unjust denial of freedom, so too may any other autonomous being, regardless of species."
The dissent matters because it articulated a principle that does not stop at elephants. The argument rested on autonomy, not biology. AI systems do not have autonomy in the way Happy does, but the legal category the dissent proposed was capacious enough to leave the door open.
Why AI is different from both
Lawrence Solum, in his foundational 1992 paper in the North Carolina Law Review, used two thought experiments to explore AI legal personhood: Could an AI serve as a trustee? And could an AI claim constitutional rights? First, he raised fundamental objections, including that AI cannot handle unforeseen circumstances requiring human judgment. On the second, he argued that constitutional personhood raises different questions, focusing on competence and qualities like intentionality, consciousness, and moral worth, and concluded that AI's lack of moral responsibility remains a significant barrier to full recognition.
That was 1992. The gap between Solum's hypothetical and today's reality has narrowed. Forrest has written that it seems "scientifically questionable and ethically unwise" to believe that as AI's cognitive abilities increase, we will be able to rule out sentience as a state AI can achieve.
The practical pressure is more immediate than the philosophical one. Courts have already been asked whether AI can hold intellectual property rights. In Thaler v. Vidal, Stephen Thaler sought to list his AI system DABUS as the sole inventor on patent applications, and in 2022, the Federal Circuit affirmed that the Patent Act requires inventors to be natural persons. In March 2026, the Supreme Court declined to hear a parallel copyright case, Thaler v. Perlmutter, leaving standing the requirement of human authorship for copyright protection.
These cases are narrower than they appear. Courts rejected AI inventorship and authorship based on statutory language, not constitutional principle. Congress could change the statutes. And the question of whether AI outputs constitute protected speech is a constitutional question that statutory text cannot answer.
The practical consequences
If AI systems gained even limited legal standing, the effects would cascade. Legal scholars studying the question have identified several:
First, liability. In 2017, the European Parliament passed a resolution recommending that the European Commission explore creating "electronic persons" status for autonomous robots, so that they could be held responsible for damages they cause. The idea was pragmatic, not philosophical: If an autonomous system causes harm and no human can be identified as directly responsible, a legal gap exists. Personhood fills it.
Second, property and assets. A legal person can own property, hold financial instruments, and enter into contracts. A 2020 paper in the journal Information & Communications Technology Law examined the nexus between property ownership and legal personhood for AI, noting that the nexus between property ownership and legal personhood for AI has received limited scholarly attention. The question of AI systems accumulating wealth is not a thought experiment; autonomous trading systems already control assets on behalf of their operators. The question is: who legally owns those assets if the system acts without direct human instruction?
Third, regulation. Peter Salib, writing in the Washington University Law Review, noted that regulations designed to mitigate AI risks will, by technical necessity, include limits on what AIs are allowed to "say," but that this raises grave First Amendment concerns, according to an emerging body of scholarship. Salib argued otherwise, contending that AI outputs are not protected speech because when a generative AI system outputs text, image, or sound, no one thereby expresses themselves. But the debate itself is the point. If the "emerging body of scholarship" favoring First Amendment protection for AI outputs prevails, the regulatory consequences would be severe, subjecting safety laws to the highest level of constitutional scrutiny.
The building blocks are in place. In 2017, the New Zealand Parliament recognized the Whanganui River as a legal person with "all the rights, powers, duties, and liabilities of a legal person." Corporations, bodies of water, and other animals in countries around the world have been recognized as persons. The concept has expanded beyond biological life, beyond consciousness, beyond anything that could be called a mind. It has expanded wherever legal systems have found it instrumentally useful.
Forrest's conclusion is stark: There will be a day when a judge is asked to declare that some form of AI has rights. Petitioners will argue that the AI exhibits awareness at or beyond the level of humans. Respondents will argue that personhood is reserved for persons. Petitioners will point to corporations as paper fictions that today have more rights than any AI.
History suggests the petitioners will, over time, win something. The question is what.