Watching the watchers — the Mother City’s expanding surveillance network demands voyeurism scrutiny

Cape Town has found a perfectly respectable excuse for a little voyeurism: public safety. On 7 October, Mayor Geordin Hill-Lewis unveiled new policing and aerial-surveillance units, promising that the City would “stop at nothing to build our resources and powers”. 

In context, the promise concerned the City’s policing ambitions generally rather than cameras in particular. Still: Nothing? That is an extraordinarily immodest promise from a government acquiring more ways to observe its citizens. One hopes there are quite a few things it will stop at. The law would be a good place to start.

A dedicated aviation unit, flying drones with thermal and zoom cameras, now joins Cape Town’s CCTV network, body-worn cameras and automatic number-plate readers. A new Joint Policing Services Operational Centre is intended to bring the feeds together. What begins as a camera overlooking an intersection can become a record of where a vehicle travelled, when it was seen and who had access to the footage. The equipment is becoming more capable; the justification for using it must keep up.

Municipal voyeurism has at least one advantage over the ordinary kind: it can be announced at a press conference. The City’s appetite for observation is beyond dispute. Whether Capetonians share it, and on what terms, is another question. Buying the binoculars does not constitute consent from everyone in view.

Yet it would be easy, and rather self-indulgent, to call the cameras Orwellian. Crime already governs the movements of countless South Africans. According to Statistics South Africa’s 2025/26 survey, only 37.9% of people aged 16 and older felt safe walking alone in their neighbourhoods after dark. That is a slight improvement on the previous year, and most household crimes the survey measures fell slightly, as did street robbery. Assault did not: victims rose from 363,000 to 408,000. 

South Africa is not experiencing a uniform rise in crime. But it does remain a country where fear imposes a remarkably effective curfew, without the inconvenience of any official having to announce one.

A commuter who avoids a route after sunset has lost a practical freedom. So has the shopkeeper trading behind reinforced barriers. The constitutional case for privacy cannot disregard the liberty already taken by criminal violence. If a camera helps make a dangerous road safe to use, it may protect freedom rather than diminish it. There is, however, the small matter of whether the cameras earn their keep.

In September, the City described an operation in Elsies River in which CCTV helped investigators identify suspects linked to smash-and-grab incidents. A drone then kept visual contact with one fleeing suspect for about 20 minutes, relaying his position until officers caught him. Five suspects were arrested. It is a convincing example of surveillance assisting a particular police operation, although an arrest establishes neither guilt nor that the wider network prevents crime.

The City also reported that its Metro Police Strategic Surveillance Unit detected 41,005 incidents in the year ending June 2026, leading to 541 arrests. Impressive numbers, until one asks what an “incident” means, how many arrests led to convictions and how many crimes were prevented.

The cost

A council can count cameras and detections with admirable precision. The crimes that never happened are harder to photograph. The cost is easier to estimate: a former councillor puts the combined value of the City’s published CCTV contracts since 2019 at R293-million or more. With local elections approaching, residents are entitled to ask what that buys.

The evidence deserves more attention than the hardware. A 2019 systematic review by Eric Piza and colleagues found modest overall crime-prevention benefits from CCTV, with stronger results in some settings, including car parks and residential areas. Active monitoring accompanied by other interventions performed better. That supports targeted use backed by effective policing. It does not establish that every additional camera is money well spent, much less that a drone will remedy Cape Town’s violent crime.

The argument becomes less comfortable when the person in the frame is attending a demonstration rather than fleeing an arrest. During the immigration protests of 30 June, officers streamed live body-camera footage to the City’s operations centre to augment drone and CCTV feeds.

Police may need such information to protect demonstrators and respond to violence. There is neither evidence in that fact alone of unlawful surveillance. Nor does the point turn on one’s view of that day’s cause: the same capability will watch any crowd. But the technology that followed a fleeing suspect in Elsies River can also record someone attending a lawful protest. What becomes of the footage when the crowd goes home? May officials identify participants, search their movements or share the recordings for some other purpose?

South Africa has reason to take these questions seriously. The apartheid Security Branch collected information about political opponents, intercepted communications and maintained files on organisations it wished to suppress. 

Today’s municipal crime prevention is not that apparatus in a new uniform. Treating it as such would insult history and abandon the argument. But history tells us something about the usefulness of dossiers to those who possess them, particularly when the people described in them have little say over their use.

Democracy offers no magical protection against official mischief. In amaBhungane v Minister of Justice (2021), the Constitutional Court considered evidence that police had obtained authorisation to intercept two journalists’ communications by falsely presenting their numbers as those of suspected ATM bombers. 

The court found serious deficiencies in the safeguards governing covert interception. Telephone interception is far more intrusive than an ordinary street camera, and the legal rules differ. The case nevertheless disposes of the soothing proposition that surveillance will remain innocent because its operators have benevolent intentions.

Protection of privacy

Our Constitution protects privacy, expression, assembly and association. An obvious reply is that a public street offers comparatively little privacy, and the Constitutional Court has some sympathy for it. In Bernstein v Bester, the court observed that the scope of personal space shrinks as a person moves into communal life. Yet the courts have also learnt to distinguish looking from watching.

In Phillips v Bradbury (2025), homeowners in Bakoven, Cape Town, defended two cameras overlooking a neighbour’s courtyard and pool as protection against crime. The Western Cape High Court dismissed their appeal against an order to reposition them. 

Less intrusive measures were available, and “systematic and continuous monitoring is significantly different from occasional observation”. Apparently, protecting one’s home does not require a view of the neighbour’s swimming pool. 

A street is not a courtyard, and municipal cameras raise a different question. But the distinction travels. A passerby glimpsed at a corner is not a person whose journeys can be reassembled from linked feeds. An observation that is fleeting in the street can become surprisingly permanent in a database. That is where the voyeurism problem lies: not in the glance, but in the album.

The Protection of Personal Information Act (Popia) matters here, though not in the way usually assumed. Consent is not its only lawful basis, and nobody should demand a warrant each time an officer looks at a traffic camera.

Adequate safeguards

The sharper point is section 6(1)(c). It excludes crime-prevention processing by public bodies only to the extent that adequate safeguards have been established in legislation. Which legislation establishes those safeguards for municipal drones, number-plate readers and a combined operations centre? Answering that means examining national and provincial law, but the City’s own rulebook offers little. Its published CCTV policy, adopted in 2014, was written mainly for privately owned cameras, lists June 2016 as its next review date, is confined to CCTV and does not mention Popia. 

A policy is not legislation. In the absence of adequate legislative safeguards, the City cannot simply assume that its crime-prevention activities fall outside Popia. Section 57 may matter too. It can require the Information Regulator’s prior authorisation where unique identifiers of data subjects are used for a purpose other than the one intended at collection and linked to information held by other responsible parties, subject to the Act’s exceptions. 

Whether number-plate data and the City’s arrangements meet those conditions is an open question, and residents cannot answer it without knowing how the operations centre actually works. At the same launch, the mayor reported a 35% increase in SAPS requests to view City footage. That shows the footage is being used more, not that it is securing more convictions. Recordings that travel between organisations need rules that travel with them.

So let the CCTV-led Cape Town make its case, system by system. Tell residents which offences each deployment is intended to reduce and publish evidence of the result, including outcomes beyond arrest statistics. Explain who may retrieve recordings, how access is audited and when information is deleted. Show that areas subjected to intensive monitoring receive meaningful protection, rather than merely excellent footage of the crimes committed there. 

Where the capacity to track people becomes more intrusive, the justification and independent scrutiny should become more demanding. The council could put all of this on a firm footing with a surveillance-technology bylaw: a public register of systems, an impact assessment before each new deployment and an annual report to the council. The US cities of Seattle and Oakland have adopted versions of the idea. A city that likes to be compared favourably to others could adopt a better one.

People who want protection from robbery should not have to apologise for welcoming an effective camera. But the City should be able to explain what its cameras prevent, what their recordings reveal and who has permission to look. For a municipality so enthusiastic about watching the public, a little public scrutiny of the watchers should hardly be considered indecent. DM