Yesterday, SecurityBrief Australia released a report that every city planner and security operator needs to read. Titled "AI arms race to reshape Australia's cyber defences by 2026," it outlines a stark reality: we are no longer just fighting human hackers. We are fighting autonomous AI agents.
The report, highlighting predictions from Kinetic IT, warns that "autonomous 'agentic' malware" is emerging - capable of scanning networks and adapting tactics on the fly. It specifically flags smart-city risk as a major vulnerability, predicting that "at least one major city could suffer a coordinated cyberattack" this year.
01 · THE HONEY-POT PROBLEMThe honey-pot problem
In this new AI arms race, the old way of building smart cities is dangerous. Traditional surveillance cameras stream massive amounts of raw footage to central servers. In 2026, that central server isn't just a storage unit - it's a honey pot. A single high-value target for the new AI attack agents.
If a security strategy relies on centralised processing, it has given the adversary a single point of failure.
02 · THE SOLUTION: RESILIENCE THROUGH EDGE COMPUTINGThe solution: resilience through edge computing
The strongest defence is decentralisation. The Occivar platform processes data at the edge. By analysing video feeds locally on the sensor hardware and only transmitting anonymous metadata - crowd density numbers, flow rates - the system effectively removes the target.
- No PII in transit. Faces never reach the cloud, because they never leave the device.
- No honey pot. There is no central database of video footage for an AI agent to breach.
- Local resilience. If the network goes down, edge units keep running on the last-known configuration.
03 · THE TAKEAWAYThe takeaway
The SecurityBrief report is right - the arms race is here. The answer isn't just "better firewalls." The answer is smarter architecture.
In 2026, privacy by design isn't just an ethical choice. It is a security imperative.