Kristin Bergtora Sandvik (2026)
A Minister Needs Help
This summer, AI glasses suddenly became a hot topic in the Norwegian public sphere. What had for years existed as a kind of utopian object of the future suddenly appeared for sale at opticians. Digitalization Minister Karianne Tung was accused of being uninformed, uninterested, and too accommodating toward the technology giants. “A capitulation to Meta,” complained a trio of influential technology commentators and practitioners in an Op. Ed piece that went viral, “It seems Tung has forgotten who she works for.”
In response to the criticism, Tung commented that it is not possible to ban every technology one dislikes. It quickly became clear that this was the wrong answer, and Tung promised to establish a task force to consider a ban.
A few short autumn weeks later, Tung was no longer Minister of Digitalization and Public Administration, and the baton passed to Torgeir Micaelsen. He immediately promised a ban but acknowledged that he was not entirely sure how such a prohibition should be designed nor which part of the supply chain it should target, or how a ban should be operationalized. Shortly thereafter, the government introduced a temporary ban on AI glasses in “selected locations.”
In August, before the semester began, I introduced a home-made ban on the use of AI glasses in all of my classrooms. I also promised that I would not use them either. As AI glasses became an issue of public debate, the normally rather technical issue of crafting regulatory bans became an important political problem. This fall, I have devoted some of my students’ time to helping Micaelsen. We have engaged in a pedagogical dugnad, derived from the old Norse dugnaðr meaning voluntary work conducted collectively.
I teach sociology of law, ethics, and robot and technology regulation at the University of Oslo. My students are Norwegian and international. Some study law or sociology of law, while others pursue a master’s degree as continuing education in downtown or are enrolled in an Honors Bachelor programme. In the Robot Regulation course, students produce films. In ethics, they meet practitioners. In the other courses, we use interactive whiteboard teaching and role-playing exercises. We have explored critical approaches to regulation and examined how particular questions come to be defined as “problems” within technology regulation and how these understandings of problems become regulation. This is an inventory of the issues we have dealt with in our dugnad efforts.
What is a ban?
While the major questions of AI regulation (Should AI be regulated? Why? Who should decide the rules, how, and when?) are important for AI glasses, the notion that only a regulatory ban is sufficient, requires some critical reflection.
A legal ban means that someone, for example the state, declares that something is not permitted. This does not mean, however, that a ban is necessarily lawful or effective. The late sociologist of law Thomas Mathiesen observed that lawyers’ affections for any kind of legal rule – including bans – meant that a “light-switch function” was almost always at play: when a legal rule is produced, lawyers expect people to familiarize themselves with it, understand it, and comply with it simply because it is, after all, a rule. Yet rules can have both intended and unintended effects. Whether a ban works depends on the social field being regulated, how the ban is articulated, communicated, and understood and what enforcement and sanctioning mechanisms exist, and, crucially, the type of resources devoted to communicating what the prohibition entails and the consequences of violating it.
Tung was onto something when she discussed the problems of prohibition as policy. Legislators struggle to keep pace with technological development, and regulation must attempt to balance innovation against rights and efficiency against the legitimacy of the law. A prohibition that suppresses innovation on paper but remains unenforced carries a double disadvantage.
As an academic topic of inquiry, technology regulation involves not only the interpretation of legal instruments such as the EU AI Act, but also the analysis of how AI glasses (or drones, data centers, and social media) operate as sociotechnical systems in which technology and society co-constitute each other and are closely intertwined.
Regulating data streams differs from regulating small physical objects. Regulating technology for industrial and commercial use differs from regulating technologies worn on the body. A prohibition may target supply chains or include restrictions on which components may be used and by whom; what functions a product is allowed to have; what market access it enjoys and which customers may purchase it; how it may be marketed and to whom; and what pricing and taxation regime applies to it, among other things. Importantly, it also requires dynamic approaches to stakeholder mapping. An interesting observation regarding AI glasses concerns the active involvement of Norwegian optical retailers. Opticians are experts in eye health, yet their websites have suddenly become full of legal information, recommendations for responsible AI-glasses behavior, and liability disclaimers. Where does this profession fit within the regulatory discussion? What kind of responsibility do they have – and should they have a voice in the regulatory conversation?
Regulation is more than law
One of the pleasures of teaching students from different disciplines is that it becomes possible to discuss forms of regulation beyond legal prohibitions. In their seminal article on regulatory challenges for robotics, Ronald Leenes and colleagues introduce a regulatory toolbox that in addition to law, also includes regulation through markets, social norms, architecture, code, and design.
When consumers no longer wanted Segways, the manufacturer stopped producing them. The popular movement to remove mobile phones from classrooms has, well, succeeded in removing them. In front of the law faculty in downtown Oslo there are large benches and heavy flowerpots. They are functional and aesthetically pleasing, but they serve another purpose as well: they are bollards making it impossible to drive a truck into the building. During King Harald’s funeral in September 2026, geofencing was used to close central Oslo to e-scooters.
I have taught Robot Regulation since 2018, and AI-regulation has been a crucial part of the course the entire time. During this time, we have moved from the flood of ethical guidelines and declarations that appeared six to eight years ago, through the EU AI Act, to today’s regulatory landscape, where the United States and China pursue distinct regulatory paths. However, over the past year, we have introduced three new concepts into our teaching: overregulation, underregulation, and deregulation.
In global conversations about tech regulation, much attention is devoted to alleged EU “overregulation” and whether deregulation is needed. At the same time, we are presently living with the security, democratic, and environmental consequences of American underregulation. Fewer rules also mean less democratic control over who governs technological development, under what conditions it occurs, and how it affects society.
What about the opticians, I wondered as I clicked through another slide.
Had I drafted this essay ten years ago, it would probably have contained a lengthy paragraph about the need for guidelines and ethical principles that could eventually lead to effective regulation. Ethics concerns doing good, but also the deliberations that precede decisions. Had I written five years ago, I would probably have clicked my tongue irritably, pounded on the keyboard, and spoken about ethics-washing and the cynical use of “ethics shopping” and ethics discourse by technology companies seeking to avoid strict regulation of AI glasses. Nevertheless, in 2026 my approach has changed yet again: ethics is central to understanding the problems associated with AI glasses, their use, and possible regulatory responses. Ethics just does not need to carry the entire burden on its own.
The Language of Technology Regulation: “Game Changers,” Luddites, and Utopians
Whether one deals with drones, e-scooters, or AI glasses, one’s fundamental perspective on technology matters. I usually present this as a continuum with gradual transitions from technological pessimists (or “Luddites,” the British weavers who protested mechanized textile production in the early nineteenth century), through sceptics and optimists, to more evangelical forms of technological utopianism.
In the latter view, technology is assumed to be capable of solving political problems, distributive and justice-related problems, and discussions often unfold in the language of technological hype, filled with phrases such as “game changer” and “revolution in …” something or other, usually in English. The focus is often on “solutionism”: the belief that social problems can be solved through technological innovation and that people are entrepreneurs capable of solving their own problems through technology.
While searching for illustrations for several lecture slides, I continued browsing the website of the aforementioned optician chain. There it was: the masculine head of a shaved-headed man, illuminated against a dark background, wearing glowing glasses. The text read: “Performance AI Glasses. GAME CHANGER.”
Regulation is also about power, ownership, and financing. Norway has long been a country of trusting and affluent technological optimism. Increasingly, however, many of us have become sceptics. The Luddites among us are no longer merely objects of ridicule accused of resisting progress. Technology prophets and IT experts have, to some extent, become scapegoats, though there is not much evidence of self-reflection. An odd convergence remains with respect to how both the most skeptical and the most enthusiastic understand technology’s effects. We call this “technological determinism”: the notion that technology functions exactly as promised and therefore brings either heaven or hell. While in the real world, things seldom go according to plan (Things break, get flooded with hot coffee, are forgotten somewhere, or are hacked), determinist visions continue to shape much of societal backdrop for regulatory debates. Yet, a sign that the technology debate has nevertheless changed is the disappearance of claims about technological neutrality, the notion that AI glasses are neither good nor bad and that their effects depend solely on how people use them.
Making AI Glasses a Problem
Having reflected on prohibition-based regulation and the language of technology, the next step is to examine how something comes to be seen as a problem that requires regulation. This does not happen automatically; problems must be produced. Carol Bacchi’s “What’s the Problem Represented to Be?” approach unpacks how policy and regulation do not merely react to problems; they actively contribute to shaping what counts as a problem. Bacchi proposes six useful questions: What is the problem represented to be? What assumptions underpin that representation? How did this representation come about? What remains silent or invisible? What consequences does the representation produce? How is it maintained and challenged?
So, what exactly is the problem here? What are the consequences? What follows from presenting and discussing AI glasses in one way rather than another? New technologies may have problematic effects. They may trigger strong objections or moral panic, as armed drones did around 2012, or become objects of ridicule, as happened with Norway’s Smittestopp contact-tracing app. In turn, societal and political responses shape regulatory debates. Both the public and political actors regularly take too long to recognize problematic aspects of new technologies, as illustrated by the case of social media and e-scooters.
AI glasses are part of a global ecosystem facing major challenges related to the energy consumption of data centers, the collection and use of AI training data, and attacks on digital infrastructure by both state and private actors. At the same time, AI glasses are a perfect object of tangible moral panic. They can be looked at, tried on, and purchased by you and me. They render the comfortably familiar in uncomfortable, ambiguous ways. Many of us wear glasses. We may also use our mobile phones to film random things in our everyday life: the sunset, a beautiful flower, a particularly funny cat. Yet this activity is generally visible to those around us.
Meanwhile, AI glasses have evolved from a distinct product category, resembling VR headsets or diving masks and featuring (allegedly) clear warning signals such as a flashing light, into devices that closely resemble ordinary eyewear. It is difficult to object to recording when one cannot know for certain that it is taking place.
The risks and benefits of AI glasses are framed in familiar ways, with the filming of children topping the list of unacceptable activities. The twist is that, in this metaphorical framing, we are all children. Organizations representing blind and visually impaired people argue that AI glasses can be valuable aids in everyday life. Yet, it is hard to deny that this is a zero-sum game: the right to use such glasses comes into conflict with other people’s right not to be subjected to surveillance, whether in swimming pools, kindergartens, or shops. Meanwhile, substantial stigma is beginning to attach itself to the technology as a public understanding of AI glasses as a form of “pervert-glasses” is starting to take hold.
Faced with AI glasses, most of us seem to feel anger, powerlessness, and indignation. The AI-glasses issue has suddenly become an issue capturing the popular imagination and thus a rewarding political cause. As I discovered while looking for illustrations for my lectures, Norwegian municipalities quickly began to respond. Across the country, local newspapers reported that municipal boards, HR-directors, and mayors were worried about the technology, that they were discussing local bans or that they had adopted emergency bans.
At the same time, as is clear that this debate involves unusual stakeholders. As I kept surfing the websites of optical researchers, I noticed substantial normative framing. Visitors are encouraged to use the glasses legally and responsibly both in daily life and at work. The retailers acknowledge that the glasses are controversial and demarcate their own role in clearly circumscribed ways: it is to explain how the glasses function, keep information updated, and provide practical advice and ethical guidance. The responsibility for data management, however, lies with Meta. While this may be legally accurate, it is an interesting communication strategy. Customers with privacy-related questions are invited to contact customer service. It is an appealing possibility that the voice answering the customers helpline now possesses specialized expertise in personal data protection law. What, I wonder, does a juridified optical industry mean for a legal-policy debate?
What Will Come of It? Why regulatory debate matters
It is a little after five o’clock on a Wednesday afternoon at the end of September, and my lecture for the Honors Bachelor students at Blindern entered its final act. I finished talking, and now the plan was to undertake a bit of dugnad for minister Micaelsen through a role-playing exercise. I divided the students into three groups. Furthest to my left sat the fictional lobbying firm “BRILLEFINT,” (sadly untranslatable – in Norwegian it is a concept combining glasses and superlatives when something is impressive) tasked with developing arguments and strategies against an AI-glasses ban. In the middle sat the equally fictional civil-society organization “NixCreepyGlasses,” which campaigns for a total prohibition. The fictional Norwegian government occupied the seats to the right.
We worked our way through strategies and proposals. In their plenary intervention, the lobbyist BRILLEFINT dodged the issue, conjured up imaginaries of costly overregulation, and talked about everything except surveillance. NixCreepyGlasses sketched out an end times scenario, arguing passionately, though full of doubt, for a ban. The government, meanwhile, deliberated carefully about the politics of the possible. Advising Micaelsen, it turned out, was not so easy. I asked the fifty or so students to call the optician’s customer service department the next day to seek answers to some of our questions.
Thursday arrived.
In the Robot Regulation class, we were doing a workshop to identify themes and groups for the student projects. Many are interested in AI glasses. A student, from one of the many countries in the world where nationalism is advancing, government is becoming increasingly authoritarian, and digital surveillance serves as a key instrument of control, approached the podium. “You know,” she said, “there was a student demonstration in my country the other day. The police used AI glasses with facial recognition to identify who took part. I want to make my project about that.”
Micaelsen, are you listening?
Acknowledgment
This essay has been translated from Norwegian by Microsoft co-pilot and revised. I am grateful to my students and to Tatanya Valland, Therese Sefton, and Maria Jumbert for comments on the Norwegian language manuscript.
Kristin Bergtora Sandvik is a Professor of Sociology of Law at the Faculty of Law, University of Oslo and a Professor II Sámi and Indigenous Peoples Law, UiT the Arctic University of Norway
Foto: University of Oslo