Twelve months ago, a tech founder named Dan Thomson declared a new country. He named it after his AI company, Sensay, claimed a small island in the Philippines' Palawan province, and installed a governing council of AI bots modeled on historical leaders: Winston Churchill, Eleanor Roosevelt, Marcus Aurelius, Nelson Mandela, Sun Tzu, Leonardo da Vinci, Alexander Hamilton, and Mahatma Gandhi. Then he opened applications for residency. About 12,000 people applied. Thomson has since admitted he is not entirely sure this will end well.
"Um, if it starts acquiring weapons and attacking neighboring islands, that would be a bad situation," he told CNN Travel, adding quickly that he considered it extremely unlikely. The reassurance was perhaps more revealing than he intended.
Sensay Island has no international legal recognition. Its current population consists of one groundskeeper named Mike. The Palawan provincial government did not respond to requests for comment on the governance arrangement. Thomson has not confirmed what documentation he holds for the lease and development rights he says he acquired. In most practical senses, Sensay Island is not yet a country.
What makes the project worth taking seriously, despite all this, is not the island. It is the 12,000 people who registered interest in living there. Thomson says he was surprised by the response. Some of those who reached out are now helping set up the project. One new collaborator, Piotr Pietruszewski-Gil, had previously tried to establish his own micronation. There appears to be a real constituency for this kind of experiment, even when the experiment is openly acknowledged to carry uncertain risks.
The micronation tradition is old and eccentric. The Principality of Sealand has operated on a decommissioned World War II naval platform off the English coast since 1967, complete with its own royal family and passports. The bohemian Republic of Uzupis in Vilnius, Lithuania, has become a tourist destination. What distinguishes the recent wave of tech-founded micronations is the ideology behind them: not whimsy, but a serious if sometimes naive belief that small-scale governance experiments can produce insights that conventional nation-states cannot.
Sensay Island belongs to this lineage but adds something new. The claim is not just that small governance is better, but that AI governance might be better than human governance. The council of historical bots is presented not as a novelty but as a feature: figures with long time horizons, no personal financial interests, and no electoral pressure. The implicit argument is that the pathologies of human politics, short-termism, tribalism, rent-seeking, might be reducible by removing humans from the decision loop entirely.
It is worth pausing on how much that argument understates the difficulty of what is being attempted. AI systems modeled on historical figures are not those figures. They are language models shaped by training data, with all the biases, gaps, and artifacts that implies. A bot trained to reason like Churchill will reproduce Churchill's documented views, including the ones we would now find indefensible, while lacking the contextual judgment that made him useful in the specific circumstances he actually faced. Governance is hard precisely because it involves navigating situations that were not in the training data.
Thomson seems to sense this, which is perhaps why his self-assessment is candid rather than boosterish. The experiment is earnest in a way that larger AI governance claims often are not, and that candor is somewhat disarming. Twelve thousand people apparently find the idea worth betting a residency application on. Whether that reflects genuine enthusiasm for AI governance, distrust of conventional states, or just the internet's appetite for novel institutions is hard to say. Possibly all three.