Transplanting biometrics in police CCTV may create Frankenstein effect

By Professor Fraser Sampson, former UK Biometrics & Surveillance Camera Commissioner

There’s something quaintly British about the fact that we still talk about ‘CCTV’.

The UK has grown up with council operated closed-circuit television cameras on our streets. We have a rough idea what they do and why they’re there. We’re comfortable with these overt surveillance systems and, like other features we drive or walk or cycle past every day, the cameras have blended into the background. They’ve become unremarkable.

However, the nature and reach of public space surveillance is changing fast – and it has nothing to do with closed circuits or televisions. This is not just another British obsession with language – as a London Borough announces that it is putting police live facial recognition (LFR) capability in its street surveillance systems, we should pause before cross-fertilizing CCTV cameras with AI-enabled biometrics.

Asked to think of ‘state surveillance’, most people picture a classic CCTV system. Cameras – usually white and invariably above head height – look down on somewhere, recording anything happening below. In this cinematic scenario, images are relayed to a central control room where people monitor banks of screens, intervening and directing as the plot dictates. Most UK streets are covered by these traditional CCTV systems, but the evolutionary direction of surveillance ecology means the future will be very different. Private space surveillance is now conducted by multi-modal devices, accessed remotely by anyone with the permissions and connectivity to do so, from anywhere in the world. The focus of AI-driven biometric surveillance will be people rather than places and it will undoubtedly replace the traditional CCTV we have grown familiar with. But we’re not there yet and we’re nowhere near ready.

Outwardly the cameras (computers) may look similar but biometric surveillance systems are a completely different species from CCTV. Behind the lens nothing is the same. Even with dynamic pan, tilt and zoom (PTZ) functionality and streaming live images, CCTV cameras are ingesters and regurgitators of images and sounds. The regulatory framework for public space surveillance in the UK is still based on this CCTV model, as is the ‘Orwellian’ vision of the dystopia we are warned about. Both are historical artefacts and fall far short of the reality ahead.

AI-driven surveillance did not evolve out of cameras on sticks. CCTV is photography, LFR is biometrics, two different disciplines processing data in very different ways – if you don’t know why, ask your national data regulator. Biometric surveillance masquerading as CCTV is not just a wolf in sheep’s clothing; it’s not even a ‘two mammal’ problem and implanting AI-enabled biometrics into CCTV risks creating a Frankenstein effect for police surveillance.

The council’s announcement comes just as the Metropolitan Police Service is beginning to recover from its early LFR setbacks (caused mainly by leaping into biometric solutions). Success has come from being careful, cautious and consultative about LFR and keeping it distinct from other public space surveillance. The results have been promising.

Crime statistics in the retail sector tell us two things about CCTV: 1) the cameras aren’t working and 2) the introduction of LFR is reducing offending in shops. Here too, confidence in the technology is growing as the early results show how it can help retailers assert control of their business space in a way that publicly funded CCTV has consistently failed to do for 50 years. It’s understandable then that local authorities might want to replicate this impact. But councils are not controlling private commercial space – justifying biometric surveillance of our streets is much more complex than stopping unwanted people coming into shops.

When you next pass a council CCTV camera remember this: all the community consultation before putting it there took place in a completely different surveillance era. So did all the research on its impact. Walk through UK streets today and the likelihood of your being identified, let alone located, simply by coming within range of one of the many closed systems of fixed-point street cameras is low to non-existent. AI-enabled biometrics like facial recognition will change that. Biometric surveillance devices aren’t just better CCTV cameras scanning a space and recording whoever passes; they are actively looking for people and they ‘know’ who those people are. The surveillance device of the future will know where I have been, where I am probably going, when and with whom. It can already teach other devices and learn beyond its original programming; it can comb and combine information in huge volumes and at extraordinary speeds. It can communicate and problem solve. These features are incomparable with the traditional closed-circuit cameras above our streets and staring up at one already tells you nothing about what it’s doing or who it’s doing it for.

The ability to recognise me instantly, match me with my passport and driver’s licence, bus dashcam or neighbour’s doorbell image, doesn’t just change the camera. Knowing the old CCTV can now do all this and may be actively trying to find me changes both me and the road I am walking, profoundly and irreversibly.

Perhaps that’s the point but we haven’t begun to understand the impact of this biometric surveillance future, let alone how it can be regulated.

There’s more to accountability than technical interoperability – and there’s much more to AI-driven biometrics than facial recognition. Get LFR wrong and the use of other biometrics for policing such as gait, heartbeat and what I’ve called zoemetrics, will be a much harder sell. Genetically engineering police biometrics into first-gen CCTV infrastructure risks a Frankenstein effect in surveillance, technologically, legally and societally. Biometric surveillance is not CCTV and needs a totally different approach from that of the street camera generation. After a few missteps the recent police case for using accountable facial recognition has been convincing and consistent – to stick the landing, it now needs coordination. 

About the author

Fraser Sampson, former UK Biometrics & Surveillance Camera Commissioner, is Professor of Governance and National Security at CENTRIC (Centre for Excellence in Terrorism, Resilience, Intelligence & Organised Crime Research) and a non-executive director at Facewatch.

Article Topics

biometric identification  |  biometrics  |  cctv  |  facial recognition  |  Fraser Sampson  |  video surveillance