Sydney’s push for AI street cameras tests trust and accountability

A simple sentence in a council paper can have far-reaching implications. Sydney’s proposal to expand its network of security cameras with artificial intelligence should prompt more than a quick technical debate about pixels and analytics. It should force a civic conversation about what kind of city we want to live in. The council’s plan — to give the camera network the capacity to “look at what’s happening on the ground much more closely” — speaks to a familiar tension. On one hand, there is the promise of safer streets. On the other, there is the risk of normalising constant automated scrutiny of everyday life. The change being proposed will not merely add new kit to lamp posts. It will change how public space is seen, catalogued and acted upon, potentially every hour of every day. For many residents, the attraction is straightforward. Technology that can flag abandoned bags, detect a scuffle or map crowd movements seems like a force-multiplier for busy police and stretched council resources. That appeal grows when headlines list another car chase, another violent incident or a spate of late‑night antisocial behaviour. But public safety is not the only outcome of more intrusive monitoring; it is also an input to how problems are defined and which responses are considered legitimate. There are well-documented downsides to AI in public surveillance even where it delivers on its promises. Machine systems are trained on data that can reflect human bias. They struggle with edge cases and can be opaque about why an alert was raised. That opacity matters: if an automated system leads to police action, who can scrutinise the decision-making chain and hold the relevant authority to account? If data is retained for long periods, who can access it and for what purposes? The council’s proposal must answer those questions before the cameras start watching more closely. Privacy is not a purely private concern. Public surveillance changes social norms. People who routinely cross town for work, for faith or for leisure have a legitimate interest in knowing whether their time in public is being recorded and analysed. Community trust is fragile; it is lost not just when the technology is abused but when decisions about its deployment are made in a technical vacuum without meaningful public input. There is a pragmatic case for clear, enforceable guardrails: narrowly defined objectives for camera analytics, strict limits on data retention, transparent reporting on how alerts are used, and independent oversight of whether the technology’s use is proportionate. Councils should be clear about who will manage the system, whether footage or metadata will be shared with other agencies, and what remedies are available if errors occur. Trials should be public, time-limited and subject to external audit. Sydney’s streets are not a laboratory in which residents must be the unwitting subjects of an experiment. If AI tools are to be part of city safety, they must be introduced in a way that strengthens rather than erodes civic trust. That means design and governance that are as public-facing as the cameras themselves. The council can have both safer streets and accountable government — but only if its decisions are accompanied by clarity, transparency and meaningful oversight. Without them, a proposal to see what is happening on the ground more closely risks encouraging the city to turn its back on the very public it aims to protect.