- SMB/SME
- Enterprise
· ASC Editorial Team
Patched but Exposed: Why Configuration Decides Whether a Flaw Becomes a Breach
In the past four weeks we've written about attacks on Palo Alto and Check Point firewalls, Fortinet's November advisories, refrigeration controllers across Europe, N-able's N-central and an AI framework that automates break-ins. Each came with the same first instruction: patch. Each also had a second lesson that mattered as much. Something that should never have been reachable from the internet was.
N-central is the clearest example. Nineteen of the 21 servers that researchers found compromised were running N-able's latest hotfix. The attack only worked because a feature, the Legacy API, was switched on and reachable from the internet. It is off by default. Someone had turned it on, and no patch could undo that decision. We covered the attacks in our N-central article.
Firewalls show the same pattern. Palo Alto rated the PHP2Shell flaw 9.2 out of 10 where a firewall's management interface could be reached from the internet, and 5.2 where access was restricted to a jump box, as we explained in our PHP2Shell article. On Check Point gateways, attackers who got in through the Mobile Access portal went on to reach the virtual machine hosts behind it over SSH, using a vCenter service account, as described in our Check Point article. Several of Fortinet's critical November flaws can be used by anyone who can reach a FortiManager's FGFM port or a FortiAnalyzer's OFTP port. For one of them there is still no fix for version 8.0, so the protection is a setting: keeping the JSON API disabled, which is the default. The details are in our Fortinet article.
It isn't only IT. The refrigeration controllers attacked across Europe in November were exposed to the internet, often running unsupported firmware. The NCSC said the attacks probably exploited known vulnerabilities in old technology. We looked at them in our article on the fridge attacks. The FBI's first recommendation on ARTEX, an AI framework attackers use to automate break-ins, is an inventory of everything an organisation has on the internet, with management interfaces reachable only from trusted networks. Among the techniques it lists are weak administrator passwords and web APIs with no authentication, as we covered in our ARTEX article.
The pattern is uncomfortable for anyone responsible for security. The tools in these incidents were firewalls, gateways and remote management platforms: the systems that are meant to defend the organisation, and the ones with the most access to it. Their defaults were often sound. The exposure came from a change made later, often for a good reason, and never looked at again.
This is not an argument against patching. Patch quickly, starting with anything that faces the internet. But a patch closes one known flaw, after it is known. Configuration decides how many of the next flaws can reach you at all.
Three checks follow.
First, know your management plane. List every administrative interface: firewalls, VPN and remote access gateways, remote management tools, hypervisors, backup systems and building or plant controllers. For each one, write down who can reach it. The answer should be trusted networks only, never the internet.
Second, treat every non-default setting as a decision with an owner. A legacy API, a remote access portal, an integration or a contractor's access should each have a reason, a named owner and a date to review it. If nobody can say why a feature is on, switch it off.
Third, check what your security tools can reach. A gateway that can open an SSH session to a hypervisor is a route in as well as a control. Review the access your firewalls, remote management tools and their service accounts have to the systems behind them, and remove what they don't need.
Which feature on your firewalls or management tools is switched on today because someone needed it once?
Get Insights by email
A short round-up of new articles every Monday morning. Nothing else.