(Image: reve.art) In 2025, my friend Cormac Cullinan, a Cape Town–based environmental attorney, won the global Shackleton Medal for the Protection of the Polar Regions. He received it for his Antarctic Rights initiative — a campaign to have Antarctica and the Southern Ocean recognised, in law, as a legal person. He has long argued (and written a book about it, Wild Law: A Manifesto for Earth Justice) that natural phenomena like the Antarctic, Table Mountain and other assorted wonders of the Earth should be given legal protection, which would give them the best chance of long-term survival against anthropogenic degradation.
Cullinan’s novel approach was to apply for a legal recognition of nature’s “inherent rights” (akin to human “personhood”) as the best mechanism to achieve environmental sustainability.
His arguments were once seen as fringe activism. Not any more. Hence, this most prestigious award.
In a bizarre twist of modernity, the same argument has now popped up in the world of AI. AI agents should be awarded legal personhood status, its proponents argue. As one can imagine, this has attracted some apoplectic pushback.
In the past few weeks, the argument has become public. On 3 June, Argentinian President Javier Milei — writing in the Financial Times alongside his deregulation minister Federico Sturzenegger, and having dispatched a draft Bill to Congress days earlier — declared that Argentina would create a new legal category, the “non-human corporation”: a company owned and operated entirely by AI agents or robots, with human shareholders permitted but not required.
It was one of three pillars, the others being a pledge to keep AI wholly unregulated and a low corporate tax rate to lure the world’s AI firms to Buenos Aires. Milei framed the non-human corporation as the natural heir to the limited liability company invented by the Dutch East India Company in 1602.
“As much as the Industrial Revolution freed us from the constraints of the human muscle,” he wrote, “AI will free us from the constraints of the human brain.”
Immediate pushback, albeit calm and respectful, came from Yuval Noah Harari, who replied in the same newspaper under the headline “We must not grant AI agents legal personhood”. Granting such status, he argued, would hand an AI an all-purpose master key to our financial, economic and political systems, with no human left to hold accountable when things go wrong — prison, the ultimate deterrent for a wayward human executive, being meaningless to a machine.
A country that does this, he warned, risks becoming something for which the historical record offers no analogy: not a company-state but an “AI-state”, a place whose inhabitants could be governed by non-human corporations. He reached, pointedly, for the same Dutch East India Company that Milei had celebrated — not as Amsterdam in its heyday but as Batavia (now Jakarta), the Dutch East India-built colonial port on the island of Java, where the company governed people for its own enrichment (and very much to the detriment of its essentially indentured locals).
Milei rebutted publicly, in a statement issued through the Office of the Presidency. Granting legal personhood, he countered, would make these entities easier, not harder, to govern — placing them inside the legal system rather than outside it, where authorities could monitor them, fine them, disable them, decertify them, seize their assets or dissolve them.
“Giving legal personhood to AI agents does not mean launching the Judgment Day of Terminator,” he wrote, invoking Asimov, and gently noting that Harari himself had praised limited liability in his book Sapiens as one of humanity’s most ingenious inventions.
(It is worthwhile to note that billionaire AI entrepreneur — and provocateur — Peter Thiel has chosen this moment to relocate to Argentina. That is not a coincidence.)
A very real question
The question of legal personhood for AI is neither academic nor philosophical. It has become very real.
How did this idea leap from the intellectual salon ruminations to a presidential Bill in the space of such a short time? For that, it is worth eavesdropping on the technologists rather than the lawyers. During the week of 6 June, Anthropic published an internal paper, When AI builds itself, reporting that more than 80% of the code merged into its own codebase was now machine-written, and that the length of task its models could complete unaided had stretched from about four minutes’ worth of skilled human work in early 2024 to the better part of a day by mid-2026 — a horizon now doubling roughly every four months.
So we have polymath technologist Alex Wissner-Gross writing tenderly that for AI agents “there will finally be a legal home for you on this planet”. Or tech investor Dave Blundin, who supplies the killer case study — the world’s largest asset manager has an AI adviser that is 99.9% accurate, yet cannot deploy it in the US for fear of being sued over the 0.1%.
Strip away the boosterism, and a sober pattern remains. AI agents are crossing the threshold from tools to actors. The binding constraint is no longer capability but permission — whether an agent may hold an account, sign a contract, move money, or be sued. Microsoft’s AI chief, Mustafa Suleyman, siding firmly with Harari, has argued that agents should have “no more rights or freedoms than a laptop computer”. It is a clean position. It is also, I suspect, already overtaken by events.
Which is why I find myself, uneasily, closer to Milei’s narrow logic than to its grand framing. Not the deregulatory zeal — that part should worry everyone — but the accountability argument. The agents are coming whether or not we name them. An unnamed agent that drains your account or forges your signature is a ghost: there is no one to summon, fine or shut down. A legal person, by contrast, can be sanctioned, even though the sanction in this case would be digital shackles and not prison walls. Read that way, personhood is not a charter of rights for the machines. It is a leash.
And here the two stories — the Antarctic and the algorithm — fold neatly into one. Cullinan’s insight was that personhood is a protection: a way of giving the voiceless standing within a system that only recognises persons. For Antarctica, the thing being protected is the wilderness. For AI, given the near-certainty of human-disintermediated agents soon running large parts of our lives, the thing being protected is us — and personhood may turn out to be our only real defence against AI fraud and deceit and malice. DM
Steven Boykey Sidley is a professor of practice at JBS, University of Johannesburg, a partner at Bridge Capital and a columnist-at-large at Daily Maverick. His new book, It’s Mine: How the Crypto Industry is Redefining Ownership, is published by Maverick451 in South Africa and Legend Times Group in the UK/EU, available now.