Visibility Was Never Consent

In R. v. Jarvis, the Supreme Court of Canada considered a teacher who had covertly filmed students in a school's classrooms and hallways using a camera hidden in a pen, in spaces where other people could see them and where security cameras already operated. The students were visible the whole time. The Court held they retained a reasonable expectation of privacy in those circumstances. It did not create a single public-space rule. It applied a contextual, multi-factor analysis that included the location, the manner and subject matter of the recording, the relationship between the people involved, the school's rules, and the recording's purpose.

That distinction is the whole argument this series has been circling for months, decided years before smart glasses existed. Being seen and being recorded are not the same event, and the difference is not sentiment. It is permanence, reach, and what a stored image can later be searched against. A courtroom already ruled that visibility alone did not decide the privacy question in that context. The industry's stock defence asks the public to forget that a court got there first.

Privacy is not the only right this touches, and treating it as the only one understates the case. The UN Human Rights Committee's guidance on peaceful assembly states plainly that gathering in public does not strip participants of privacy, and names facial recognition specifically as a way that right can be infringed. A protest steward worrying about who is filming the crowd is not being precious about etiquette. They are responding to a chilling effect this series has already measured behaviourally, and international human rights law has already given that effect a name that sits outside data protection entirely: interference with assembly.

Children sit in a third category again, not a smaller version of the adult one. The Convention on the Rights of the Child protects children from arbitrary interference with their privacy in their own capacity, and UN guidance on digital environments extends that explicitly to connected devices and biometric tools. A child who can point at a camera and understand it is recording still cannot reasonably be expected to grasp what happens to that footage five years and one recognition service later. Handing a ten-year-old the burden of saying no is not a safeguard. It is a redirection of liability toward the person least equipped to carry it.

None of this requires waiting for legislation modelled on any of it. A design question can stand in for all three bodies of law at once, and we would put it in front of every product decision this category makes: what can the person in frame perceive before or during capture, what can they do about it without escalating the encounter, and what happens when they have no realistic way to leave.

Answer those honestly for a classroom, a clinic waiting room, or a protest, and most of what ships today fails on the third question alone. A recording light answers the first, poorly. Almost nothing on the market today answers the second or third at all, because both were treated as somebody else's problem, usually the bystander's, to solve with their own composure in the moment.

Every argument we have made about firmware, liability, and market discipline assumes the underlying premise is still contested. It may not be, or at least a court has already ruled on a version of it. The premise the industry keeps re-litigating in public, that visibility settles the question by itself, already failed as a complete answer in a case about a different camera in a different decade. Nothing about smart glasses makes that reasoning newly correct. It just makes the reasoning newly profitable to repeat.