Larry Ellison, the Oracle billionaire who counts the Trump administration among his closest clients, recently said out loud that citizens will be on their "best behavior" because AI systems will be "constantly recording and reporting" everything they do. He said it like it was a selling point. According to security expert Bruce Schneier and surveillance researcher Jon Penney, writing in The Guardian, he's right about the technology and catastrophically wrong about why that's acceptable.
What This Technology Actually Does
We are not talking about some dystopian future scenario here. The infrastructure is being built right now. Schneier and Penney describe systems that combine real-time facial recognition, AI-powered analysis, mass databases, and personalized enforcement into something that functions like an automated cop who never sleeps, never forgets, and has no concept of proportionality.
Think speed cameras, but they cover every surface of public life. Shoplifting, littering, jaywalking, attending the wrong protest. The system sees it, logs it to your government record, and notifies authorities in real time. The Guardian piece frames it plainly: you will not get a ticket in the mail three weeks later. You will be flagged immediately. The machinery of consequence will activate before you have finished the act.
China is the working prototype. The country has over 600 million surveillance cameras, with AI and facial recognition increasingly powering enforcement of both legal and social rules. The authors cite the case of Lao Duan, a Chinese citizen who fell behind on loan repayments and was blacklisted. When he later visited Beijing, the city's surveillance system identified his face at a major intersection and displayed his photo, name, and citizen ID on a nearby electronic billboard with the message that he was an "untrustworthy person." Public shaming as automated infrastructure. That is the model.
America Is Not Watching This From a Safe Distance
Here is where the piece lands hardest. The Guardian reports that the US Department of Homeland Security is rapidly expanding its own AI-based surveillance operation, including facial recognition tools and monitoring of social media accounts, to track immigrants, dissidents, journalists, legal observers, and protesters. That list is not incidental. Those are not random categories of people. Those are specifically the people whose scrutiny of government power is supposed to be protected by the Constitution.
And Ellison's quote about citizens being on their "best behavior" is not an offhand remark from some random tech executive. Oracle works closely with the Trump administration. This is a man with enormous contracts and enormous influence saying, without embarrassment, that the goal is behavioral compliance. Not safety. Not crime reduction. Compliance.
The surveillance experimentation is global, according to Schneier and Penney, with systems being tested across North America, South America, Europe, Asia, and Africa. But the United States under the current administration is not a passive observer of this trend. It is an active participant with enormous resources and, increasingly, an ideological interest in the outcome.
The Chilling Effect Is Not a Side Effect. It's the Product.
Penney has a new book on exactly this subject, titled Chilling Effects: Repression, Conformity, and Power in the Digital Age, and the core argument he and Schneier make is that the most dangerous thing about AI surveillance is not the individual enforcement action. It is the aggregate change in how people behave when they believe they are always being watched.
Surveillance causes self-censorship. It makes people more conformist. It makes dissent feel dangerous and creativity feel risky. And the authors argue these effects are additive: the more mechanisms you stack together, the more powerful the chill. What AI does is fuse all of those mechanisms into one persistent, unrelenting system with no off switch and no forgetting.
This is not theoretical. The piece draws a clear historical line: the FBI's domestic surveillance program in the 1950s and 1960s, which used wiretaps, mail opening, informants, and paper index cards to track alleged communists, looks genuinely primitive compared to what is being built now. East Germany's Stasi, which built the most extensive human-surveillance network in history, relied on people. Fallible, gossip-prone, occasionally sympathetic people. The new systems have none of those weaknesses.
What Gets Killed When Everybody Knows They're Being Watched
Schneier and Penney make an argument that deserves more attention than it usually gets in these conversations. Social progress requires a period of illegality. The normalization of same-sex relationships and legal marijuana did not happen because legislators woke up one morning with good values. It happened because there were communities of people willing to live differently, publicly and privately, in defiance of laws that were wrong. They demonstrated, over decades, that the moral consensus was movable.
If AI surveillance existed at scale during those decades, would that counterculture have been possible? The authors think probably not. When every act of social deviance is logged and penalized in real time, the cost of being ahead of the moral curve becomes prohibitive. The experimentation that drives social change gets priced out of existence.
That is not a hypothetical concern about some future generation. It is a direct argument that the systems being built right now will lock in the current moral consensus and make it much harder to ever change. Whatever you think about where society is today, the idea that it should never be permitted to evolve further is terrifying.
So What Can Actually Be Done
Schneier and Penney are not fatalists, and to their credit they do not end the piece with a shrug. They argue that bans on facial recognition and other identification technologies can slow development. Robust data protection laws can limit what gets collected and stored. Transparency and accountability requirements can at least force these systems into the open where they can be challenged.
The key word in all of that is "can." None of it happens automatically. None of it happens because technology companies decide to be reasonable. It happens because democratic governments make deliberate policy choices to reject the surveillance infrastructure being offered to them by people like Larry Ellison, who has already told you exactly what he thinks the point of it is.
The policy window is not permanently open. China built its surveillance state over years while the policy debates lagged badly behind the construction. The United States appears to be on a similar trajectory, with the added complication that the current administration shows no particular interest in the kinds of civil liberties constraints that might slow any of this down.
The Dingo Take
Let's be honest about what Larry Ellison said. He did not accidentally reveal the agenda. He stated it, confidently, as a feature. Citizens will behave because they know they are being watched. That is not a description of public safety. That is a description of a controlled population. And he said it while his company holds contracts with the administration currently using DHS to monitor journalists and legal observers.
The comparison to China's billboard shaming system should not feel as distant as it does. The technical difference between displaying Lao Duan's face on a Beijing intersection screen and sending an automated alert to a law enforcement database when your face appears at a political rally is a difference of degree, not kind. The underlying logic is identical. You are being tracked. Your record follows you. Your behavior will be shaped accordingly.
Schneer and Penney are right that this is not inevitable. But "not inevitable" is doing a lot of work in a political environment where the administration's billionaire allies are openly celebrating the chilling effects as the whole damn point. The time to make different policy choices is before the infrastructure is complete, not after everyone has already learned to stay home.
