Your Office Has 50+ Smart Cameras. New UAE Research Shows How Fast They Can Be Hacked.

This week, at the 12th World Internet of Things Conference at Khalifa University in Abu Dhabi (October 26–29), an Emirati researcher presented an AI model that detects cyberattacks on smart-home devices, like cameras, dramatically faster than before. The story being reported is a defensive breakthrough. The story underneath it is the one worth paying attention to: someone had to build this because small, networked devices like cameras are already a live target — and most businesses aren't running one or two of them. They're running dozens.

What Was Actually Announced

Mohammed Al Shehhi, a data scientist and lecturer at the Higher Colleges of Technology, presented a new detection model built specifically for resource-constrained IoT devices — the low-power chips found inside things like home security cameras. Reports put its accuracy at 97.45% across 14 different attack types, and roughly 31 times faster than the prior international benchmark tested on the same device. The model uses explainable AI to isolate the network-traffic signals that actually matter, which is what keeps it light enough to run on hardware with limited processing power in the first place.

In other words: the device being defended is small and unremarkable on purpose, because that's exactly the kind of hardware attackers are already going after.

Why This Matters Past the Smart Home

The research was tested against a single home camera. A typical office, retail outlet, warehouse or residential building isn't running one — it's running a CCTV network, access control panels, intercoms, gate barriers and visitor kiosks, often dozens to hundreds of connected endpoints. Every one of them is a smaller, cheaper version of the exact hardware this research set out to protect.

Attackers don't need to breach a core server if an exposed camera or access panel hands them a foothold on the same network. And unlike a single home device, most installed commercial security hardware ships with factory-default credentials, rarely receives a firmware update after install day, and frequently sits on the same flat network as everything else — because to whoever set it up, “it's just a camera.”

What Actually Protects a Network of Cameras and Access Control

None of this means you need a research-grade AI detection model of your own. It means the basics most installations skip are the ones that matter most:

• Network segmentation — keeping cameras, access control and other IoT devices on their own VLAN, separate from core business systems, so one compromised device can’t move freely across the network
• Credential and firmware management — default passwords changed at install, firmware patched on a schedule rather than left exactly as it shipped
• Centralized monitoring — one platform watching every connected device for unusual behavior, instead of each camera or panel being its own blind spot
• Site-specific design — placement, network load and integration planned around the actual site and device count, not a generic package rolled out everywhere the same way

Closing the Gap With the Right Partner

This is exactly where CODE3 designs security in from the start rather than bolting it on afterward. Our CCTV and AI Camera installations, Access Control & Biometric Attendance systems and Network Infrastructure are planned together — segmented networks, managed credentials and centralized monitoring — for businesses across Dubai, Abu Dhabi and Sharjah. Both sit alongside our broader cybersecurity services, scoped to what your site actually runs rather than a one-size-fits-all package.

UAE researchers are actively working to defend devices like these faster. Talk to us about getting a quote to make sure the cameras and access panels protecting your business aren’t the gap attackers find first.