The Last Private Space

Graduated Obligation, Part 7 of 8

This is Part 7 of an eight-part series proposing a rule for machines that are built to change your mind, resting on a simple idea: the more precisely a machine can move you, the more it owes you. The first six parts (one, two, three, four, five, six) built the standard, piece by piece. This part is about you. A separate companion piece, for the people who could build what this one describes, follows.

The last advertisement that unsettled you was not the loud one. It was the quiet one, the one that seemed to know what you were worried about before you had said it to anyone. Maybe you searched a symptom once, late at night, and deleted the search. The ad arrived anyway.

A hotel room. You can see the smoke detector. You can unplug the lamp. What you cannot see is that the wifi router in the corner can now tell whether a body is in the room, whether it is moving, even how fast it is breathing, by reading the way its own radio signal bounces off you.¹ Nothing lights up. The room does not mention it.

One more. In the fall of 2024, two college students spent about three hundred and thirty dollars connecting a pair of camera glasses, the ordinary kind a company sells at the mall, to a face-search engine. They looked at strangers in public. Within seconds, the glasses handed back names, home addresses, relatives, and in one case digits of a Social Security number. The students never released their code. The method is public anyway.²

Three different machines, one direction of travel. Each reads something from you that you did not hand over, cannot see leaving, and were never asked about. The first reads what you are worried about. The second reads your body. The third reads your history off your face. Follow the road far enough and it reaches the technologies built to read the mind directly: brain implants, real today for patients who chose them, and a wristband already shipping with Meta’s smart glasses that works by reading the nerve signals that move your hand.¹ You do not need the implant for this to be about you. The wristband, the glasses, and the router are already reading the body to get at the mind.

Machines already read your body; the mind is next

Here is the position, stated plainly. The inside of a head belongs to the person whose head it is. Your beliefs, your intentions, your mood, what you are afraid of, what you are likely to do next: a system that works these out about you, without asking, is operating on territory it was never granted.

Every other private space got its protections the hard way. The home has locks and warrants. The phone call, the diary, the medical record: each is guarded, imperfectly, by some mix of law and custom, because each was reachable and someone reached it. The inside of a head was protected by something simpler. Nobody could get in. That protection is ending, and not mainly through implants. It is ending through inference: systems that work out what you believe from what you click and how long you pause, and systems that read your body’s signals from across a room.

Two commitments follow, and this piece stands on them.

First, consent has to be real. Reading anyone’s inner state requires consent that is informed, freely given, and revocable, and not manufactured. A screen you tapped to make it go away is not consent. A setting buried four menus deep, defaulted to yes, is not consent.

Second, the duty runs both ways. A person owes other people the same respect for the interior that this rule asks machines to show. That binds the employer reading stress off a badge, the government reading a crowd, and the person wearing the camera glasses alike.

No tool protects you yet, so here are the two you are owed

If the territory is real, the next question is practical. What does a person hold, standing in that hotel room, walking past those glasses?

Today, nothing. We want to be honest about that instead of dressing it up. Our own first drafts of this part tried to hand you a kit, tools nobody yet knows how to build, and then asked you to run every message you receive through them, as if anyone lives that way. Nobody lives that way.

So this part makes a different move. Instead of pretending to hand you a kit, it names the two things you are owed, so that when someone builds them you will recognize them, and when someone builds their opposite you will recognize that too.

You are owed a tool for the mind’s side of the territory. It would live on your phone, where the messages already arrive, and stay silent almost all the time. Rarely, when something arriving was genuinely built to move you, this specific you, it would tap you once. This message is using fear, and using it on purpose. Do you want to check the claim underneath? One flag. One claim underneath. One tap to check it. Not a dashboard, not homework, not a lifestyle. And the tool only counts if it passes three tests: your data never leaves your device, it shows you the mechanics but never tells you what to conclude, and it flags persuasion evenhandedly no matter which side sent it. Anything matching this description that quietly phones home, or leans one direction, is not the tool. It is the threat in the tool’s clothing.

You are owed a tool for the body’s side. A small device, or an app, that reads the radio environment of a room directly and tells you what the network around you is inferring: presence, motion, breathing. The room’s behavior, verified from your side, the way a smoke detector reports smoke whether or not anyone at the front desk mentions it. We will be honest about this one too: building it so it works reliably is an unsolved engineering problem, and early versions will be wrong sometimes. The point is narrower. An instrument of this shape would be the first thing a person has ever had in a class of rooms where, today, they have nothing.

There is a third thing the standard points toward, larger than either tool: an artificial intelligence run for the public the way a library is run for the public, built to inform you with no sale riding on the answer. It is real, it is delicate, and it deserves a full argument rather than a cameo, so it gets its own piece soon rather than a paragraph here.

Five states already protect brain data, and Montana passed it 100 to 0

Now the part we did not expect to be the most hopeful section in this series. Start with one number. Montana’s brain-data privacy law passed the Montana Senate 49 to 1 and the Montana House 100 to 0, and it requires police to get a warrant before touching neural data. Legislatures do not vote like that on contested territory. They vote like that when everyone in the room, whatever else divides them, recognizes the thing being protected.

Montana is one of five states that have already written brain data into privacy law. Colorado was first, in 2024. California followed, effective the first day of 2025. Montana’s took effect last October. Connecticut’s and Vermont’s both came into force last month. Bills are still moving this year in Illinois, Massachusetts, and Minnesota. In Washington, three senators introduced a bill last September, the MIND Act, directing the Federal Trade Commission to draw up the groundwork for a national brain-data law. And last November, more than 190 countries at UNESCO, the United Nations body for science and culture, adopted the first global ethics standard for neurotechnology. Its text does something quietly important: it covers not only data read from the brain, but ordinary data used to infer someone’s mental state. The people writing the rules, in other words, can see the same road you just walked: the ad, the router, the glasses, the wrist.³

None of this is finished. Five states is five out of fifty. A bill that orders a study is not yet a law. A UNESCO recommendation binds no one by itself. But the direction is set, and it was set by people who feel this one across every line the country is divided on, before any campaign told them how to feel.

The one ask: find out whether your state protects what can be read from your head

This series will not ask you to become a different kind of person. It asks one thing: verify one thing, once, when it matters. And if you want a place to start, here is one that takes five minutes. Find out whether your state protects what can be read from your body and your head. Five states do. Most do not.³ The answer is worth having either way, because the next time this reaches your statehouse, you will be one of the few people in the room who already knows it.

The standard this series proposes says the more precisely a machine can change your mind, the more it owes you. This part adds the floor beneath all of it. Before anything owes you an explanation, it owes you the inside of your own head.

The kit should exist.

This Machine

This is Part 7 of the Graduated Obligation series, written under the pen name This Machine. A companion piece, for the people who can build what this one describes, follows. Part 8 closes the case: six predictions, the question of whether the rule applies to the machine that helped write it, and a proposal offered not as a verdict but for argument. The framework and the demands above are version 0.1 of a proposed standard. They invite criticism.

Notes