Wearable Surveillance Is Wrong Before It Is Illegal
Day 247: Fear Friday Arachnephobia by Kathryn
I asked a question in an online thread about a pair of AI camera glasses, and the most useful answers I got were all about the law. One reader pointed out that in public there is no reasonable expectation of privacy, so there may be no legal issue at all. Another explained that in the EU you can record others under the household exemption for personal use, but that the exemption narrows once you record at volume or for non-personal purposes, at which point you may become a data controller with obligations attached. A third, writing from Spain, noted that he has the right not to be filmed by a private doorbell in a public place, yet anyone he speaks to may record the conversation, because the law there permits one party consent.
I had asked whether it was legal or ethical for my daughter to be recorded by a stranger's glasses without her consent. The thread answered the first half and left the second half sitting there, which is the half I actually care about. My daughter is what made me notice the missing piece. This new technology should not be viewed as ‘normal’ and we should not assume that public life has always worked this way.
The phrase "no expectation of privacy in public" was built for a world in which being in public meant being seen by whoever happened to be present, in the moment, and then forgotten. A stranger watching you cross a street is observation. It is embodied, bounded, and reciprocal, since you can see them seeing you. A wearable camera actually captures, and the moment it is recorded, it’s retained, and routed to infrastructure you have no relationship with, where it can be matched, analyzed, and inferred upon long after you have walked away.
Nissenbaum's account of contextual integrity explains that privacy, in her framing, is not secrecy but the expectation that information moves according to the norms of the context it was shared in. Being visible to the people in a restaurant is one context. Becoming a frame in a stranger's video that uploads to a corporate model is a different one. The law collapses the two because it still measures privacy by whether information was exposed, not by where it travels afterward. Solove (2025), in the Florida Law Review, makes the structural version of the same point: privacy law is built almost entirely around the input, the act of collection, and is close to blind to the output, the inference and generation that follow. The recording is legal. What the recording becomes is ungoverned.
The Consent Argument Turns Against the Wearer
The clearest current example is Meta's camera glasses. A February 2026 New York Times investigation followed what happens when they enter ordinary physical spaces. The recording indicator is a small light most people miss or mistake for Bluetooth. One creator interviewed for the piece described approaching people who had no idea they were being filmed. The device introduces what is best called ambient collection. The wearer captures not only their own surroundings but the faces, voices, and reactions of everyone in range, none of whom agreed to anything.
This is not ordinary photography. Magee, Ienca, and Farahany, writing in Neuron, define cognitive biometric data as physiological and behavioral signals that can be processed to infer mental states, and their review of seventeen companies found that nearly all collect it in some form. A face in a frame is no longer just a face. Combined with expression, gaze, and context, it becomes what Heller terms biometric psychography, the extraction of a psychological profile from physical signals. The bystander did not consent to being photographed. They certainly did not consent to being read.
Here the consent question turns, and it turns against the person holding the camera, not the person in front of it. When I accept a platform's terms, I am exercising a thin kind of self-determination, technically bargaining away my own data, badly and asymmetrically, but it is mine to bargain. Solove calls digital consent a legal fiction, "murky consent," so far removed from real understanding that it barely describes a decision at all. Even granting that fiction, it only covers me. The person my glasses record signed nothing. The household exemption that several readers invoked was written for a photo album and a home movie. It was not written for a device that records continuously and sends what it sees to an inference engine.
There is a cleaner test than arguing jurisdiction by jurisdiction. My favorite law philosopher from university, John Rawls, proposed judging the fairness of an arrangement from behind a veil of ignorance, without knowing which position in it you will occupy. Let’s apply it here.
Not knowing whether you will be the one wearing the glasses or the one walking past them, the one capturing or the one captured, would you agree to a world in which any stranger may record you at any time and route the result to a private data center?
Almost no one would. The arrangement looks acceptable only from the position of the person holding the camera. That is usually the sign that it is not a fair arrangement at all.
What a Watched Population Becomes
The individual violation is real, but it is not the deepest problem. The deeper problem is what a population becomes when it assumes it is always being recorded. And there is evidence for this, not only intuition. Penney (2016), in the Berkeley Technology Law Journal, documented a measurable drop in traffic to sensitive Wikipedia articles after people learned the extent of online surveillance. Awareness alone, with no enforcement, was enough to suppress ordinary curiosity. His later work extends the finding from state surveillance to the softer, ambient kind and reaches the same place: surveillance produces self-censorship, and over time, conformity. Foucault described the mechanism half a century ago when he envision the effectiveness of the panopticon was that it needed no guard in the tower. The mere possibility of being watched made the watched begin to police themselves. The gaze gets internalized. People perform compliance whether or not anyone is looking.
Scale that across a public where a meaningful share of people wear cameras, and the chilling effect stops being situational and becomes the weather. You no longer adjust your behavior for a particular watcher. You adjust it for the standing possibility of any watcher. Véliz (2020) argues that privacy is not a private preference but a collective form of power, and that ceding it hands the holder the capacity to shape a society. A watched population is an agreeable one. It says the safe thing, holds the expected view, and keeps the unfinished thought to itself. That is the passivity worth worrying about, and it exists without anyone deciding to impose it.
What makes this version harder than the surveillance state is that there is no state in it. No government agency built it. No legislature authorized it. It is assembled from the bottom up by consumers who bought a product, and it is owned by the companies that sold it. There is no tower to petition and no single guard to remove, because the guards are each other, and most of them believe they are only filming their lunch. A panopticon you can vote against is one kind of problem but a panopticon you are helping to build, one purchase at a time, is a harder one.
The readers in that thread were probably right that the courts will eventually reach some of this. Cases will come, definitions will be tested, and a few of the worst uses will be named as illegal. But law is slow and behavior is fast, and by the time the rules arrive the population will already have adjusted to being recorded. A right you assert only after the architecture is finished is a right you have already half surrendered.
So the question the law keeps asking, whether you have a reasonable expectation of privacy in public, is not the one that matters. My concern is if a society, in which everyone assumes they are being recorded, can still think out loud. Cognitive privacy is the freedom to wonder, question, and form ideas without those processes being observed. We used to be able to carry that freedom outside with us. The last unwatched space was the physical one and it’s now being closed by people who mean nothing by it, and that is precisely why it is so hard to defend.
References
Foucault, M. (1977). Discipline and punish: The birth of the prison (A. Sheridan, Trans.). Pantheon Books. (Original work published 1975)
Heller, B. (2020). Watching androids dream of electric sheep: Immersive technology, biometric psychography, and the law. Vanderbilt Journal of Entertainment and Technology Law, 23(1).
Magee, P., Ienca, M., & Farahany, N. A. (2024). Beyond neural data: Cognitive biometrics and mental privacy. Neuron, 112(18), 3017–3028. https://doi.org/10.1016/j.neuron.2024.09.004
Nissenbaum, H. (2010). Privacy in context: Technology, policy, and the integrity of social life. Stanford University Press.
Penney, J. W. (2016). Chilling effects: Online surveillance and Wikipedia use. Berkeley Technology Law Journal, 31(1), 117–182. https://doi.org/10.15779/Z38SS13
Penney, J. W. (2022). Understanding chilling effects. Minnesota Law Review, 106(3), 1451–1532. https://doi.org/10.24926/265535.4359
Solove, D. J. (2024). Murky consent: An approach to the fictions of consent in privacy law. Boston University Law Review, 104, 593. https://ssrn.com/abstract=4333743
Solove, D. J. (2025). Artificial intelligence and privacy. Florida Law Review, 77(1), 1–85. https://ssrn.com/abstract=4713111
Véliz, C. (2020). Privacy is power: Why and how you should take back control of your data. Bantam Press.
Timothy Cook is Director of The Cognitive Privacy Project and author of the "Algorithmic Mind" column at Psychology Today. He is Securiti Certified in AI Security & Governance.
Contact: timothy@cognitiveprivacyproject.org Web: cognitiveprivacyproject.org
© 2026 Timothy Cook / The Cognitive Privacy Project. All rights reserved.Licensed under CC BY-NC-ND 4.0. You may share this work with attribution. Commercial use and derivatives require written permission.
The public conversation frames AI as a tool that can be used responsibly. But in environments that mediate cognitive work, AI acts as an invasive species. It outcompetes and displaces the native, effortful processes required for human reasoning. This ecological framing demands a fundamentally different approach to AI governance.