Smart glasses for police officers sound like one of those ideas that is either going to make policing considerably more effective or give us a very uncomfortable glimpse of what happens when technology gets a little too enthusiastic about watching everything. The basic idea is pretty straightforward: equip officers with glasses that can record what they see, display information in their field of vision, potentially identify people or objects, translate speech or text, provide navigation, access databases, and perhaps use AI to help an officer make better sense of what is happening in front of them. On paper, there is a lot to like about that, particularly when an officer is dealing with a fast-moving situation where having to look down at a phone or radio is simply not practical.
The good bit is fairly obvious because if an officer’s glasses can automatically record an encounter, that creates a contemporaneous record of what happened rather than relying entirely on someone’s recollection afterwards. We already know body-worn cameras can provide useful evidence, but smart glasses could potentially take that a step further because the camera is actually looking in the same direction as the officer. There is also the possibility of giving officers information without making them take their eyes off what is happening. Imagine an officer arriving at an incident and being able to see the location of the call, relevant warnings, a translated message, a building layout or information about a vulnerable person without rummaging around for a phone or trying to read a screen while standing in the middle of a developing situation. For policing, where seconds can matter, that could be genuinely useful rather than merely shiny technology looking for a job.
There are also some interesting possibilities around accessibility and communication. Real-time translation could potentially help when an officer and member of the public do not share a language, while speech-to-text could assist officers who need information displayed rather than communicated over a noisy radio. An officer could potentially receive directions while moving through an unfamiliar building, identify a particular address more quickly or have important information displayed while keeping both hands free. There is even a legitimate argument that recording encounters could protect the public as much as it protects the police, because there should be a record of what was said and done rather than two competing versions of events.
And then we get to the bit where my eyebrow starts heading upwards and I pull a face… because the moment you put a camera on somebody’s face and connect it to databases and AI, you have moved considerably beyond a pair of clever specs; you have potentially created a mobile surveillance platform which happens to have a police officer underneath it and I think that distinction matters. A body camera recording an incident is one thing; glasses capable of continuously seeing everybody around the officer, analysing faces, identifying objects, listening to conversations and pulling information from databases is quite another. The technology may be capable of doing it, but that does not automatically mean that the police should be allowed to do it.
Facial recognition is where this gets potentially murky; if these glasses can look at somebody and immediately compare their face against a police database, we are no longer simply talking about recording evidence. We are talking about identifying people in real time as an officer walks down the street and that raises a very different set of questions about privacy, proportionality, accuracy and oversight. What happens when the system gets it wrong? Because AI does not become infallible merely because someone has attached it to a pair of glasses. If the system incorrectly tells an officer that somebody is a wanted person, what happens next? Does the officer trust the information because it appeared neatly in their field of vision? Does the technology become an unofficial authority figure simply because it looks confident?
There is another problem which is much less glamorous but potentially just as important: what exactly is being recorded, when is it being recorded and who gets to see it? If an officer walks into a hospital, a school, someone’s home or a highly sensitive situation, do the glasses continue recording? If they automatically record, can the officer stop them? If they can be switched off, who decides when that is appropriate? If the system stores everything temporarily and only retains particular incidents, who makes that decision and how? If recordings are uploaded automatically, where are they stored, for how long and who has access? And if someone hacks the system, congratulations, we have potentially created a very impressive collection of police video footage for somebody else to use for their benefits.
There is also the human factor, which tends to get forgotten whenever somebody puts the words “AI-powered” in front of a proposal. An officer may start to rely on what the glasses tell them rather than what they can actually see, hear and assess themselves. If an AI system says a person is potentially dangerous, does that subtly alter the officer’s behaviour before they have even spoken to them? If the software highlights somebody as a “person of interest”, does the officer unconsciously treat that person differently? We can talk about algorithms being objective until the cows come home, but the data they are trained on, the databases they access and the assumptions built into the software all come from somewhere; databases themselves can be corrupted, updates may be missed or incomplete and many other permutations that make the information given to an officer false or incorrect.
Then there is the really ugly possibility: mission creep. Something introduced because it is useful for recording assaults could eventually be used for facial recognition which can then become linked to wider intelligence databases. Those databases could potentially incorporate information from multiple sources. Before long, something introduced as “a useful tool for officers dealing with dangerous situations” has quietly become a system capable of identifying and tracking people as they move through public spaces. Nobody necessarily sat down one morning and decided to build a mass-surveillance system; it just happened one additional function at a time because, well, the technology could do it.
And I think that is the bit we should be particularly wary of, because “can we?” and “should we?” are two completely different questions. Technology companies are naturally interested in what their hardware can do because that is rather conveniently what they sell, while police forces are interested in tools that make officers’ jobs easier and potentially safer. Neither of those interests should automatically determine what the public is expected to accept. There needs to be a very clear line around what the glasses are permitted to do, what they are prohibited from doing, when recording is mandatory, when it must stop, what information can be displayed, what databases can be accessed, how long data is retained and who independently checks that the rules are actually being followed.
There is also the question of whether officers themselves would want all of this. Being permanently equipped with technology that records, analyses and potentially evaluates what is happening around you sounds useful until you realise that the same technology could also record the officer making a mistake, saying something inappropriate, entering somewhere they shouldn’t or simply doing their job badly. Personally, I think that is actually one of the more interesting aspects, because proper recording should work both ways. If smart glasses are sold as a way of improving accountability, they should not mysteriously become accountability devices only when the footage is useful to one side.
And then there is another longer-term question which I think deserves considerably more attention than it probably gets when everyone is busy talking about the clever bits: what happens when we make exposure to yet another wireless device part of somebody’s job and tell them they have to wear it for years? Smart glasses are likely to contain wireless technologies such as Bluetooth, Wi-Fi or other radiofrequency transmitters, meaning that an officer could potentially have a transmitting device sitting against or very close to the head for an entire working shift, day after day, year after year. I am not saying that this means smart glasses cause cancer, because the evidence simply does not allow us to make that claim, but neither is it accurate to pretend there is absolutely nothing to investigate. The World Health Organization’s cancer research agency, IARC, classified radiofrequency electromagnetic fields as “possibly carcinogenic to humans” (Group 2B) in 2011, following an assessment by an international working group of 31 scientists from 14 countries, with the classification based particularly on limited evidence concerning glioma and acoustic neuroma in relation to wireless-phone exposure. IARC specifically said that the available evidence was limited and that chance, bias and confounding could not be ruled out, so this is not proof that RF exposure causes cancer, but it is also not a declaration of absolute safety. The current ICNIRP guidelines set exposure limits for radiofrequency fields covering technologies including Bluetooth, Wi-Fi, mobile phones and 5G, and are designed to protect against established adverse health effects, while acknowledging the scientific literature on both short- and long-term exposure.
The problem for police officers is therefore not that we can point to a study and say, “these glasses will give you cancer”, because we cannot, but that compulsory occupational exposure is a different question from somebody choosing to use a wireless device occasionally. If we are going to make officers wear transmitting technology against their heads for potentially decades of their working lives, surely the sensible approach is to establish exactly what frequencies, transmission powers, duty cycles and exposure levels the equipment produces, what happens when several wireless systems are operating simultaneously, and whether the cumulative exposure has been properly assessed over a working lifetime rather than simply ticking a box because each individual device falls within an existing limit. We have a rather bad habit with new technology of discovering twenty years later that “probably fine” was not quite the same thing as “we actually know this is safe”, and when the technology is compulsory rather than optional, the burden of proving that the occupational exposure is justified should surely be higher. There is a reason, after all, that hospital staff who work around X-rays routinely wear radiation-monitoring badges so their occupational exposure can be measured over time; if we are going to require police officers to spend their working lives wearing wireless technology against their heads, perhaps we should apply the same basic principle of measuring exposure rather than simply assuming that staying below a regulatory limit means there is nothing left to monitor.
So, yes, I can see the good in smart glasses for policing. Give an officer genuinely useful information without forcing them to stare at a handset, improve communication, record encounters accurately, assist in dangerous environments and potentially provide better evidence, and there is a very reasonable case for exploring the technology. But the phrase “smart glasses” makes it sound like we are talking about a slightly more sophisticated version of Google Glass when what we could actually be talking about is a wearable computer with a camera, microphone, AI, biometric identification and access to highly sensitive information strapped to someone’s face.
That is where I would want the brakes applied rather firmly – not because the technology is inherently bad, but because the consequences of getting the rules wrong are considerably bigger than the consequences of buying a pair of useless glasses. If the glasses help an officer see more clearly, communicate more effectively and record what actually happens, brilliant. If they quietly turn every police officer into a walking facial-recognition camera with access to a giant database of everybody they happen to look at, then we’ve crossed a line which needs considerably more public debate than a shiny technology demonstration at a police conference.
The technology is probably coming whether we are ready for it or not. The important question is whether the rules arrive before the glasses do.
