First published October 2019. Rewritten and last reviewed 12 August 2026.
We are asked to look at more existing camera systems than we are asked to design new ones. The fault is rarely the camera. It is what the camera is plugged into, and who can reach it from outside.
A camera is a computer with a lens. It has an operating system, a password, a network address and, in most cases, a path to the internet. Whether it keeps you safe depends on those four things far more than on the number of megapixels printed on the box.
What has changed since we first wrote this
When this article first went up in 2019 the common failure was a recorder bought online, left on its factory password and opened to the internet through a port on the router so that an app could reach it. Whole websites existed that did nothing but list cameras left in that state, sorted by country.
The law has since caught up with part of the problem. The UK’s product security regime under the Product Security and Telecommunications Infrastructure Act came into force on 29 April 2024. A consumer connectable product sold in the UK can no longer ship with a universal default password, the manufacturer has to publish how to report a security flaw, and it has to state for how long the product will receive security updates.
That helps with what you buy from now on. It does nothing for the recorder that has been in the cupboard under the stairs since 2017, and it does not stop anyone forwarding a port to a camera that was compliant on the day it was installed.
Why an old recorder is a liability rather than a saving
We are often asked to “fix” a system rather than replace it. Sometimes that is the right call. Often it is not, for reasons that have nothing to do with picture quality.
- It has stopped receiving updates. A recorder or camera that no longer gets firmware is a device with known flaws that will never be closed. Check the manufacturer’s published support period before spending money on it.
- It is reachable from the internet. Port forwarding and UPnP were the normal way to get remote viewing working a decade ago. Both expose the recorder’s login page to anyone who scans for it, and scanning is automated and constant.
- Nobody knows who has the app. Installers, previous owners, a relative who set it up. Each is a set of credentials that still works.
- The footage does not do the job. A camera that shows a shape at the gate but cannot identify a face is a deterrent at best. Identification is a matter of lens, position and light, and it is decided at design, not at the recorder.
How we build a camera system
The approach is the same in a private house and in a restaurant. The scale changes; the principles do not.
One cable per camera, and it is the right cable. Every camera runs on its own Category 6 or 6A cable back to a managed switch, which powers it over that same cable. There is no local power supply at the camera, no wireless link to drop out and no junction box on the wall. The cable routes are agreed with the architect at RIBA Stage 2 and 3 and drawn at Stage 4, so they are in the walls before the plaster is, and the camera sits exactly where the sightline needs it.
The cameras live on their own network. They sit on a separate VLAN with no route to the rest of the house or business. A camera cannot see the laptop, the phone or the office file store, and a compromised camera cannot be used as a way in to any of them.
Nothing is exposed to the internet. The recorder is on site and is not reachable from outside. Remote viewing is done through a VPN into the property, or through the manufacturer’s authenticated relay service where we have reviewed it. There is no forwarded port and UPnP is switched off at the router.
Every device has its own credentials, and the client holds them. Unique passwords per device, an administrator account that belongs to the owner, and named accounts for anyone else who needs to view footage. When someone leaves, their account is removed. Our own access is agreed in writing and can be withdrawn.
Updates are a maintenance item, not an afterthought. Firmware on cameras, recorder and switch is reviewed on a schedule as part of a support agreement, and the manufacturer’s support period is checked before a product is specified so that the system is not obsolete before the warranty runs out.
The law applies to the footage too. A domestic system that records beyond your own boundary, a shared driveway or the pavement for instance, brings data protection law into play, and the Information Commissioner’s Office publishes guidance on what that means for a household. For a business the obligations are wider. Camera positions and retention periods are decided with that in mind, not left to the default setting.
Five things to check on your own system tonight
- Log in to the recorder. If the password is the one it came with, or the one written on the sticker, change it now.
- Find the date of the last firmware update on the recorder and on one camera. If you cannot find one, or it is years old, the system is running on known flaws.
- Look in the router for port forwarding rules and UPnP. If either points at the recorder, remote viewing is working the dangerous way.
- List everyone who has ever had the app. If you cannot, assume more people can see your footage than you would like.
- Stand at your front gate at night and ask someone to look at the live picture. If they could not describe your face, the camera is a deterrent, not evidence.
Where we come in
Security cameras are one part of the security and CCTV work we design, and they only work as well as the network infrastructure beneath them. We normally work for the client, with the architect, so that cable routes, camera positions and the network they sit on are settled on the drawings rather than improvised on site.
If you have a system you are not sure about, ask us to look at it. We will tell you plainly whether it is worth keeping.