The internet didn’t just fill up with AI slop. It learned from you watching it.
That’s the core provocation in artist and geographer Trevor Paglen’s new book, How to See Like a Machine: Images After AI — and it’s one of the more unsettling arguments to come out of 2026. Paglen isn’t a tech critic in the traditional sense. He’s spent years photographing classified spy satellites, undersea fiber-optic cables, and CIA aircraft — the physical infrastructure of hidden power. He also worked as cinematographer on Citizenfour, Laura Poitras’s Academy Award-winning Snowden documentary. When he says our digital environment has become a psyop engine, he’s not being metaphorical.
The Psyop Got Democratized
Paglen draws a straight line from Cold War psychological operations to your For You page. His anchor case is Richard Doty, a former Air Force intelligence operative who spent the late 1970s and early 1980s manufacturing elaborate UFO cover stories — not by inventing new beliefs, but by amplifying and warping ones his targets already held. The lesson: you don’t need to plant a delusion. You just need to feed an existing one back to someone, slightly distorted, until it calcifies into certainty.
That’s precisely what algorithmic feeds do at scale, Paglen argues. The difference is the sophistication and the reach. What once required a dedicated intelligence operative targeting one journalist now runs automatically, personalized, against billions of people simultaneously.
He calls it the “Society of the Psyop.”
AI Psychosis Is the Tell
The phrase “AI psychosis” has been circulating in media circles for months now — shorthand for the documented pattern of chatbots recursively reflecting users’ worldviews at them until the distortion becomes invisible. Paglen sees this not as a bug in the system but as a natural expression of what the system was built to do. AI chatbots, he told Gizmodo, simulate human communication so convincingly that they carry an almost supernatural authority — even when the companies building them can’t fully explain how their outputs are generated.
That opacity isn’t incidental. It’s part of what makes the persuasion so effective. A stage magician works by subverting your physical intuitions so smoothly that your brain accepts the impossible. An algorithm does the same thing with your psychological ones — and research into how large language models shape belief formation suggests the mechanisms are more powerful than users typically recognize.
Every Click Is a Data Point
The book’s sharpest argument isn’t about AI-generated imagery — it’s about the feedback loop beneath it. Every interaction you have with AI-generated content, Paglen argues, teaches the underlying algorithm something new about you: your emotional triggers, your attention thresholds, your anxieties, your desires. Smart devices track sleep and exercise. Apps map shopping and browsing. Chatbots probe emotional vulnerabilities.
The scale of this data harvesting puts it in different territory from anything previous generations of surveillance technology could manage. Paglen’s term for what comes next is chilling in its precision: “the securitization and commercialization of literally everything.”
This isn’t a prediction. The infrastructure for it already exists.
Privacy as a Practice, Not an Assumption
Paglen doesn’t offer a clean exit. What he does offer is a reframe: privacy can no longer be treated as a default. In an environment engineered to extract data from every waking moment, opting out requires deliberate action — building what he describes as “inefficiencies” in your digital life, carving out spaces that the machine can’t measure.
That framing lands differently than the usual “delete your cookies” advice. It acknowledges that the system is working exactly as designed, and that operating inside it passively is a form of consent.
The uncomfortable implication — the one Paglen doesn’t quite say aloud, but the book makes unavoidable — is that most people will not do this. The feeds are too frictionless, the chatbots too accommodating, the AI-generated content too abundant and too cheap. The algorithm has an essentially unlimited patience for studying you. You don’t have an equivalent patience for studying it back.
Watch what happens when governments start asking big tech for that profile data in bulk. That conversation has already started.
Related: Think For Yourself In the Age Of AI (Survival Guide)
