From the recent CCTV User Group conference in Northamptonshire, here’s a digest of a panel of the day’s speakers, chaired by User Group deputy chair Ilker Dervish.
Professional Security asked about how CCTV managers might get hold of Safer Streets grants off the Home Office. Although, was central Government after ‘quick wins’ politically, rather than dealing with actual issues, such as crumbling CCTV equipment, someone on the panel asked. Politics does indeed come into it; the funding cycle – Safer Streets has reached round five – under any new Government after an imminent general election may offer a more open way of gaining access to the Fund – and some ‘serious money’, six figure sums to renew CCTV systems – rather than money funnelled through police and crime commissioners. Ilker mentioned the recent Home Office research on Safer Streets rounds two and three, which found that the grants made no significant impact on crime, such as VAWG (violence against women and girls), either in terms of recorded crime or feelings of community safety. The criminologist Prof Emmeline Taylor remarked on the short-termism of the funding.
That begged the question of the purpose of public space CCTV; what’s the data being gathered for, and who for? Elected councillors and ‘influencers’? Criminal prosecutions? If public space cameras are to serve the police, what of the perennial phenomenon raised by the info-security consultant Mike Gillespie of Advent IM, that police will ask for footage, walk off with it and the provider never hears of any results? Mike argued that we need to be smarter about how we use technology. Tony Porter the former surveillance camera commissioner now chief privacy officer for Corsight, a facial recognition product company, gave the view (as a former senior police officer besides) that CCTV data ‘is extremely valuable’ but was surprised that there’s not the evidence base to support that.
In answer to a question from the floor about what trends the panel sees, Tony Porter said ‘we are in the foothills of AI’. He suggested it’s inevitable that facial recognition will be used for access control; marketing analysis in shops; and on the streets, ‘to put the fear of God into people planning hostile reconnaissance is quite a big deal’. Mike Gillespie warned that video surveillance will be ‘tainted’ if AI is used to create very believable deep-takes: “This is a challenge that we should lay down to the whole of the manufacturing community; sort this out once and for all.”
In his earlier talk, Tony Porter – beginning with a compliment to council CCTV as the ‘fourth or fifth emergency service’ – stressed use of facial recognition that’s ethical, transparent, and lawful. For the British standards body the BSI he’s just finished writing an ethical approach to facial recognition; part of global work that will affect UK CCTV users. The European Union meanwhile has passed an AI Act that will have bearing on the implementation of AI (as during the summer Olympics in Paris). That said, regulations and the law cannot keep up with the tech. He recalled a commitment in the Conservative manifesto of 2019 to a law governing such tech, not yet seen in 2024: “I find that quite remarkable.”
Tony Porter and other speakers made plain that ever more augmented camera surveillance – whether DNA databases or automatic number plate recognition – is not only a technical matter of how to do something, but it’s about civics, what sort of state citizens want (and Tony noted that he was speaking on St George’s Day). To return to Tony, he said: “Let me proclaim, privacy is not dead.” The debate has become polarised, he commented, and users have to tread carefully, and do due diligence when procuring (do cameras meet international standards, such as ISO 42001, an AI management standard? As for Corsight AI’s facial recognition, he recalled a tech firm put his company through an AI panel for eight hours). If you are bringing AI into the operations room, you should have a ‘human in the loop’, Tony said. Because humans make the settings of such software, that comes up with a 70, 80 or 90 per cent similarity between the person captured on camera and the watch-list or database. In other words you should have ‘dual recognition’, such as an officer on the street besides confirming that the match.
Photo by Mark Rowe; the Lilin stand on the CCTV User Group exhibition show floor.
More in the June print edition of Professional Security Magazine.
