Two prominent technology leaders are now publicly making the same argument: if cameras and artificial intelligence tracked people constantly, everyone would act better because they'd know they were being watched.

Peter Diamandis, who runs the XPRIZE Foundation, made this case recently. He was echoing an idea that Oracle co-founder Larry Ellison floated in 2024, when Ellison described a future where police officers and ordinary citizens would both be under continuous monitoring. Ellison framed this as beneficial, saying it would keep people "on their best behavior."

The idea itself is not new. It goes back centuries to a philosopher named Jeremy Bentham and his concept of a prison-like building called the panopticon — a structure designed so that an observer could watch everyone without anyone knowing exactly when they were being watched. More recently, criminology researchers have studied whether people actually do behave better when they think someone is looking.

What is new is the technology. Artificial intelligence systems can now process video feeds, facial recognition data, and behavioral patterns in seconds, across entire cities. The cost has dropped enough that this kind of monitoring is starting to look practical at a city-wide scale. Ellison and Diamandis are suggesting that modern AI finally makes the old behavioral theory workable.

This idea deserves careful thought. The claim rests on an assumption about how humans work — that we are rational, that we think ahead about consequences, and that we change our behavior when we know we're being watched. That's true sometimes, but not always. There's also a bigger question these two men haven't fully addressed: who decides what "best behavior" means, and who gets to control the system once it's built.

The real-world picture of surveillance is already well documented. Cities like London have CCTV cameras everywhere. Police departments across the United States have tested facial recognition software. China uses both at scale. And in all those cases, problems have emerged — cameras are sometimes turned on people exercising their right to protest, facial recognition makes mistakes more often when scanning people of color, and there's no good way to check if an algorithm made an error before someone's life is disrupted.

It matters that both Ellison and Diamandis have direct stakes in this argument. Oracle sells the software and cloud systems that governments and companies use to store and analyze huge amounts of data. Ellison profits when surveillance systems become normal. Diamandis has spent decades pushing the idea that technology can solve big problems. Both are genuinely believers in what technology can do. But that belief shapes which problems they talk about and which they leave out of the conversation.

The regulatory situation varies sharply by country. The European Union has created a rule — the EU AI Act — that classifies continuous facial recognition in public spaces as high-risk and bans most of it. The United States has no equivalent federal rule. This matters because it means the debate Ellison and Diamandis are pushing will play out differently depending on where you live — and that what happens in Europe may not happen in America.

The core question is not whether AI-powered surveillance is technically possible. It is. The real question is what rules and safeguards get put in place before it becomes normal — and whether the people talking loudest about its benefits have actually thought hard about who controls it, and what that power could be used for down the road.