Working remotely with RDP? How to do it safely
Using Remote Desktop to work remotely? Why it shouldn't face the internet, what cyber insurers ask about it, and how to set it up safely.
Someone needed to reach an office PC from home, so years ago a port was forwarded on the router. It still works, and nobody has looked at it since. Working remotely over RDP isn't the problem; that one rule is. It is among the first things a cyber underwriter asks about, and among the first things an attacker looks for.
This article is for IT leads who have to answer the question on a proposal form, and for brokers who have to explain why it is there. It covers how to keep working remotely over RDP without leaving it open to the internet: what "exposed RDP" means, what insurers ask, how to find it and what to put in its place.
What "exposed RDP" means
Remote Desktop Protocol (RDP) is Microsoft's protocol for connecting to a Windows PC or server remotely. By default it listens on TCP port 3389 (Microsoft Learn).
RDP is "exposed" when the service answers directly from the public internet. The usual causes are a port-forwarding rule on the office router, or a cloud server with a public IP and no filtering. Anyone, anywhere, can then reach the login screen and try passwords.
Microsoft's own instructions for port forwarding carry a warning: "You're opening your PC up to the internet, which isn't recommended." The same page says a VPN is the preferable route.
Why it matters now
Attackers increasingly log in rather than break in. The Sophos 2025 Active Adversary Report, based on more than 400 incident-response and managed-detection cases from 2024, found:
- In 56% of cases, initial access came through external remote services, such as firewalls and VPNs, using valid accounts.
- RDP was involved in 84% of cases, making it the most frequently abused Microsoft tool. That figure covers any use of RDP during an attack, including movement inside the network, not only entry from outside.
- Compromised credentials were the top root cause, at 41% of cases.
One of the report's recommendations is three words long: "Close exposed RDP ports."
The US agency CISA says the same in its #StopRansomware Guide: "Do not expose services, such as remote desktop protocol, on the web." Where exposure can't be avoided, it asks for compensating controls.
What insurers ask about RDP
Two public proposal forms show how the question is put.
- Beazley, applicants under £20m revenue. The cyber application lists, among its core cybersecurity controls, a yes/no statement that the applicant does not expose RDP "or other remote desktop software" directly to the internet. Applicants above £1m revenue are also asked whether MFA is required for all users' remote access to the network.
- Chubb, New Zealand edition, revenue $50m to $700m. The Cyber ERM proposal form asks the applicant to describe its use of RDP: not used, used for remote access, internal only, or something else. If it is used at all, the applicant ticks which protections are in place, including VPN, MFA and Network Level Authentication.
Forms differ by insurer and market, and we found no Greek insurer's public form to compare. The logic is the same everywhere: unprotected, internet-facing RDP signals higher risk. Depending on the insurer, that can mean a higher premium, conditions before binding, or a decline.
It is also a question where the answer can be checked. An exposed remote desktop service is visible from outside, which is why it sits high on the list of outside-in signals underwriters look at before they quote. A "Yes, we don't expose it" that turns out to be wrong is a problem at claim time, not just at quote time.
How to find out whether you have it
1. Take an inventory
Ask your IT team or provider three questions:
- Which PCs and servers accept Remote Desktop connections?
- Is there a port-forwarding rule on the router or firewall to port 3389, or to another port that leads to RDP?
- Do we run cloud servers, such as Windows VMs, with a public IP and RDP open?
2. Check from outside
An internal review is not enough, because routers keep rules that people have forgotten. You need a check from the internet towards your public IP addresses. Run it only against addresses and domains you own.
3. Don't rely on a changed port
Moving RDP from 3389 to another port doesn't hide it. Internet-wide scanners look at every port, not just the defaults. A non-standard port cuts noise in your logs. It doesn't cut the risk.
How to close it, step by step
| Step | What to do | Who |
|---|---|---|
| 1 | List who really needs remote access, and to what | Owner / IT |
| 2 | Set up a VPN with MFA, or an RD Gateway, for those users | IT / provider |
| 3 | Test that users can connect through the new route | IT |
| 4 | Delete the port-forwarding rule to RDP on the router or firewall | IT |
| 5 | Turn RDP off on machines that don't need it | IT |
| 6 | Repeat the outside check and keep dated evidence | IT |
Do steps 2 and 3 before step 4. Closing the port first is how people end up locked out, and how the rule gets quietly reopened a week later.
What to put in its place
- VPN with MFA. The user connects to the VPN first and then to the PC, as if on the office network, so the PC never faces the public internet. CISA's guide asks for MFA on all VPN connections.
- RD Gateway. A Windows Server role that Microsoft describes as enabling "secure, encrypted connections" to Remote Desktop resources over the internet (Microsoft Learn). Pair it with MFA.
- IP allow-listing. If something must stay exposed, allow only specific source addresses. Beazley's form includes this among its optional controls. Treat it as a stopgap, not the destination.
Controls to keep anyway
For any RDP that remains, CISA's guide recommends account lockout after a set number of failed attempts, MFA, logging of login attempts, and Remote Credential Guard or restricted admin mode for administrator sessions.
Common mistakes
- "We have a strong password." Attackers often use credentials that have already leaked, which is why they were the top root cause in the Sophos data.
- Closing RDP but leaving the VPN without MFA. The risk has moved to a different door.
- Forgotten rules. A previous technician, the accounting-software vendor or another supplier may have opened access nobody remembers.
- Remote support tools. The Beazley statement covers other remote desktop software too. List what your suppliers use to reach you.
- Answering the form from memory. Confirm the position first, then answer.
Check what is visible from outside
Velontech's VelonRadar checks daily what is publicly visible for your domain: email authentication, HTTPS, end-of-life software and, once you have verified that the domain is yours, exposed risky services such as RDP. It reads public information only and does not enter your network. VelonGrade turns that into an A–F grade with the fixes that raise it. We're in private beta: request a free check and see where you stand before the insurer asks.
Want us to check this for you?
A free check using public information only. We reply within one working day.