A surprising number of serious intrusions begin with something that could be reached from the internet but never needed to be. It might be the administration page of a firewall, a remote desktop service opened for convenience, or a piece of equipment that nobody thinks of as IT at all, such as a building controller or a refrigeration system. None of these is exotic. They are found by attackers who scan the whole internet continuously, and who need only one of them to be weak.
This guide looks at how to keep that exposure under control. It is built around three requirements from the RiskWitness Blueprint, which work best when they’re treated as a set: two decide what you allow to be reachable, and the third checks that what is really reachable matches what you decided.
The internet
- Automated scanning looking for admin pages, remote desktop and unpatched devices
- Staff and suppliers working remotely
- Your external check: what really answers from outside? RW-NET-18
The edge
- One managed gateway, fully patched, with multi-factor sign-in: the only way in RW-NET-04
- Firewall and router admin pages: not reachable from the internet RW-NET-03
Your network
- Internal systems, reached remotely only through the gateway
- A management network or protected admin devices: the only place admin pages are used from
- A forgotten server or controller, found by the external check and closed RW-NET-18
- The one intended way in
- Closed to the internet
- Exposure found and closed
- Attackers
About the RiskWitness Blueprint
The RiskWitness Blueprint is a free information security framework that we publish for organisations of every size. It sets out 249 requirements across a core and eleven areas, from identity and access to suppliers and artificial intelligence. Each requirement says what to do and why it matters, and leaves the choice of how to the organisation.
Requirements sit in three tiers, and each tier includes everything below it:
- Foundation: what every organisation should have, whatever its size, including the controls that stop the most common attacks.
- Established: a managed programme, with deeper risk work, supplier assurance, testing, monitoring and regular review.
- Advanced: for organisations with more at stake, with round-the-clock monitoring and deeper, independent testing.
The Blueprint is also proportionate. It says what to achieve, not how far to take it, so a small firm and a large one can both meet the same requirement in ways that suit their size and risks. The full framework can be downloaded free.
What the requirements ask
RW-NET-03Foundation
Network equipment admin pages kept off the internet
Make the admin pages of firewalls, routers and other network equipment unreachable from the internet, unless reaching them also needs a second factor, and manage the equipment only over encrypted connections.
RW-NET-04Foundation
Remote access through one managed gateway
Allow remote access to internal systems only through a managed gateway that receives every security update, never by opening services such as remote desktop directly to the internet.
RW-NET-18Established
Everything exposed to the internet is known
Check regularly what your organisation exposes to the internet, and close anything found that isn't on your list of services meant to be reachable.
The first two are Foundation requirements because the exposures they deal with are among the most commonly exploited. The third is Established because it needs a running process: a list of what should be reachable, and regular checks against it. Without it, though, the first two depend on everyone remembering every change, which in practice nobody does.
Why they matter
Equipment at the edge is a favourite target
The devices that sit between an internal network and the internet are known as edge devices. In February 2025, cyber security agencies from nine countries, led by the Australian Signals Directorate, published joint guidance on securing them. It describes the most commonly targeted as enterprise routers, firewalls and VPN concentrators. The agencies give several reasons for the attention these devices receive: they sit on the perimeter, so compromising one can open a path into the internal network; newly published flaws in them are frequently exploited; and both skilled and unskilled attackers scan internet-facing services looking for vulnerable ones. State-sponsored groups have also used such devices to keep long-term access to critical infrastructure.
The guidance sets out seven mitigation strategies. One of them is direct: management interfaces should not be directly accessible from the internet. Others are closely related, including knowing where your network perimeter is and what sits on it, replacing devices that have reached the end of their supported life, and using strong, phishing-resistant multi-factor authentication.
An admin page is the keys to the device
A management interface controls everything that passes through the device, so whoever reaches it and gets past its sign-in can change the device’s rules, read its configuration or use it as a foothold. If the page can be reached from the internet, an attacker can try passwords at will, or wait for the next published flaw in that product and use it before a patch is applied.
The US government took this seriously enough to make it mandatory for federal agencies. CISA’s Binding Operational Directive 23-02, issued in June 2023, covers the remote management interfaces of routers, switches, firewalls, VPN concentrators, proxies and load balancers, as well as out-of-band server management interfaces such as iLO and iDRAC. Agencies must either remove those interfaces from the internet, or protect them with access controls enforced by something separate from the interface itself. The directive applies only to US federal agencies, but it is a clear statement of what good practice looks like for everyone else.
Remote access is the other common way in
Remote access is necessary for most organisations, which is exactly why it attracts attackers. In its 2021 annual review, the UK’s National Cyber Security Centre described the main ways ransomware groups gained initial access. Insecure remote desktop configurations, with credentials collected through phishing, breaches or credential harvesting, were one. Exploiting vulnerabilities in VPN software was another. Sending every remote connection through one managed gateway, kept fully up to date, gives attackers a single well-defended entrance to aim at, rather than several that are each looked after less carefully.
Exposure creeps
The third requirement exists because the exposure an organisation intends and the exposure it actually has drift apart over time. A supplier opens a port for a support visit and it is never closed. A test server is left running after a project ends. A new device arrives with remote management switched on by default. Equipment outside IT’s usual view comes with a web interface of its own. In October and November 2026, for example, attackers changed the settings of internet-connected refrigeration controllers in supermarkets in several European countries; France’s national cyber security agency pointed to exposed administration interfaces and weak passwords (our news report has the detail). Comparing what is really reachable with what is meant to be is the only reliable way to find this kind of drift before an attacker does.
How to meet them
How far an organisation takes these requirements depends on its size, the information it holds and the risks it faces.
A small organisation
Typical set-up
A single internet router, cloud email and file storage, and an IT supplier who supports its computers remotely.
What it could look like
- Confirm the router’s remote management setting is switched off, and that its default password has been changed.
- Agree with the IT supplier that remote support uses one named tool, with multi-factor sign-in, and that nothing else is left open for them.
- A few times a year, have the supplier or an external service check which of your addresses and ports respond from the internet, and ask about anything unexpected.
A larger organisation
Typical set-up
Many network devices across several sites, cloud networks, more than one remote access route, and suppliers with their own connections.
What it could look like
- Manage network equipment only from a separate management network or protected administration devices, so its admin pages are never reachable from outside.
- One remote access gateway, patched promptly and protected with phishing-resistant multi-factor sign-in, for staff and suppliers alike.
- A maintained list of every service meant to be reachable from the internet, each with an owner and a reason.
- Regular external scans of every address and domain you use, compared against that list, with anything unexpected closed or explained and recorded.
How to show it’s done
Leadership, auditors and customers will want more than an assurance that things are probably fine. Useful evidence includes:
- A current list of the services meant to be reachable from the internet, each with an owner and a reason for being exposed.
- The date and result of the most recent external check, and what was done about anything it found.
- A periodic, recorded confirmation that administration access and remote access are set up as the requirements describe.
- For anything that has to remain exposed for now, a recorded decision on the risk, with who made it and when it will be looked at again.
Where it usually falls short
Equipment that isn't seen as IT
Supplier access
“Temporary” changes
Checking only from the inside
Old equipment
How RiskWitness helps
RiskWitness is a hosted information security risk register. It ships with the Blueprint built in, with starter controls already mapped to its requirements, and it can be used to manage these three requirements from start to finish. The Starter plan includes the Blueprint’s Foundation tier; Team and above include all three tiers, which RW-NET-18 needs.
Record what's exposed
Add each internet-facing device and service as an asset, with the person responsible for it and where it’s kept or hosted. Link the suppliers who can reach it, so supplier access is visible alongside the asset.Record the risk
Add a risk for administration pages or remote access being reachable from the internet, with an owner. Score it before and after controls, and choose how it will be treated.Make the checks controls, and test them
Add controls such as “Admin pages unreachable from the internet”, “All remote access through the gateway” and “Quarterly external check”. Set how often each should be tested, and record every test with its result, so a failed check is visible rather than forgotten.Map them to the Blueprint
Link each control to RW-NET-03, RW-NET-04 and RW-NET-18. One control can count towards several requirements, and RiskWitness shows how well each requirement is met.Give each gap an owner
When a check finds something unexpected, create an action with an owner and a due date. Overdue actions show on the dashboard.Record exceptions properly
If something has to stay exposed for now, record the risk acceptance with a reason and an expiry date. Outside your tolerance, a named approver signs it off, and the acceptance lapses on its own if the risk gets worse or a linked control fails a test.Keep leadership informed
Produce a PDF report or slide deck covering your risks, which controls are being tested on schedule and which tests are overdue, and the actions still open.
Every step is kept in a history that shows who did what, when, and the old and new values, and nobody can edit or delete it. When leadership, an auditor, a customer or a regulator asks what you expose to the internet and how you know, the answer is already written down.
Sources
- Mitigation strategies for edge devices: executive guidance, Australian Signals Directorate’s Australian Cyber Security Centre and partner agencies, February 2025
- Joint guidance: security for edge devices, NCSC New Zealand, 5 February 2025
- Binding Operational Directive 23-02: mitigating the risk from internet-exposed management interfaces, WaterISAC, 13 June 2023
- Ransomware threat methodology, NCSC Annual Review 2021
- Defrost and disrupt: an assessment of Europe’s refrigeration attacks, SentinelLABS, 3 November 2026, updated 4 November 2026
