TRANSKRIP 🔗
WEBVTT
00:00:02.000 --> 00:00:09.679
Good morning, good afternoon, or good evening, whenever you may be, and welcome back to another episode of the SOC Brief.
00:00:09.839 --> 00:00:16.640
I'm your host Andrew, and today we're talking about an actively exploited zero day affecting Fortinet Fortamail.
00:00:16.719 --> 00:00:23.600
And normally when we hear critical vulnerability in an internet-facing appliance, the response is pretty predictable.
00:00:23.839 --> 00:00:29.039
Figure out whether we have it, find the version, patch it, close the ticket, and move on.
00:00:29.280 --> 00:00:33.679
But this one gives us a really good example of why that isn't always enough.
00:00:33.920 --> 00:00:44.320
Because we now have a real-world victim that says attackers exploited a zero day in Fortinet technology and quietly copied incoming email for more than two months.
00:00:44.560 --> 00:01:01.920
So today we're going to talk about CVE 2026-104286, what happened to Belgian network provider Bellnet, what Fortamail administrators need to do right now, and probably the most important lesson, remediation and incident response are not the same thing.
00:01:02.240 --> 00:01:04.719
So let's start with the vulnerability itself.
00:01:04.959 --> 00:01:10.640
On October 1st, Fortinet disclosed CVE 2026-104286.
00:01:10.959 --> 00:01:16.000
It's a critical vulnerability in Fortamail with a C VSS score of 9.8.
00:01:16.239 --> 00:01:20.640
Fortamail, for anyone who hasn't worked with it, is an email security gateway.
00:01:20.799 --> 00:01:30.560
It sits in the email path and is designed to inspect messages for things like spam, phishing, malware, and other threats before those messages reach users.
00:01:30.719 --> 00:01:35.760
In other words, this is literally a security product designed to protect your email.
00:01:35.920 --> 00:01:39.359
And that's part of what makes this vulnerability so interesting.
00:01:39.519 --> 00:01:44.159
The flaw involves path traversal and improper handling of null characters.
00:01:44.319 --> 00:01:51.599
And although you don't need to remember that vulnerability specifically, what's important is what it allows an attacker to do.
00:01:51.840 --> 00:02:02.239
An unauthenticated attacker can send specially crafted HTTP or HTTPS requests and potentially write arbitrary files into the underlying Fort Mail system.
00:02:02.640 --> 00:02:09.599
No valid administrator account required, no employee clicking a phishing email, no stolen password.
00:02:09.840 --> 00:02:16.879
If the vulnerable interface is reachable and the system is vulnerable, the attacker may be able to interact directly with the appliance.
00:02:17.120 --> 00:02:23.439
An arbitrary file write on a security appliance is obviously a very bad primitive to give an attacker.
00:02:23.680 --> 00:02:32.879
Fortinet says exploitation can ultimately result in unauthorized code or command execution, and this wasn't theoretical when Fortinet disclosed it.
00:02:33.039 --> 00:02:36.000
Load full transcript
The vulnerability was already being exploited.
00:02:36.240 --> 00:02:45.520
CISA added it to the known exploited vulnerabilities catalog the same day, but here's where this becomes more than another vulnerability of the week story.
00:02:45.759 --> 00:02:56.319
A few days before Fortinet publicly disclosed the CBE, Belgium's national research and education network, Bellnet, announced that it had discovered a security incident.
00:02:56.560 --> 00:03:04.240
Bellnet provides networking and related services to Belgian research institutions, universities, and government organizations.
00:03:04.560 --> 00:03:09.680
On September 24th, they detected unauthorized activity in their infrastructure.
00:03:09.840 --> 00:03:17.120
They later confirmed that the incident involved exploitation of a zero day vulnerability and technology supplied by Fortinet.
00:03:17.360 --> 00:03:20.560
And the attackers apparently hadn't just gone in that week.
00:03:20.800 --> 00:03:28.400
Bellnet determined that incoming email had been copied and transferred to external infrastructure beginning on july twenty second.
00:03:28.560 --> 00:03:34.560
They contained the incident on the morning of September twenty fifth, so we're talking about more than two months.
00:03:34.719 --> 00:03:39.199
And during that period, Bellnet says copies of incoming emails were being taken.
00:03:39.439 --> 00:03:45.759
That wasn't just metadata, the actual messages, their contents and attachments.
00:03:46.000 --> 00:03:53.599
That's a very different type of compromise from somebody defacing a server or deploying ransomware immediately announcing their presence.
00:03:53.840 --> 00:03:55.039
This was quiet.
00:03:55.199 --> 00:04:03.840
And if you're an attacker conducting espionage or intelligence collection, compromising the email gateway is an incredibly valuable place to be.
00:04:04.080 --> 00:04:06.719
Think about what travels through email.
00:04:07.039 --> 00:04:26.319
It can be things like contracts or invoices, legal documents, password reset messages, internal conversations, incident response discussions, customer information, business plans, configuration details, attachments, potentially credentials or API keys that somebody absolutely should not have emailed.
00:04:26.480 --> 00:04:29.279
But we all know somebody eventually will.
00:04:29.519 --> 00:04:33.920
And there's an interesting security lesson hidden inside the architecture here.
00:04:34.160 --> 00:04:42.079
Organizations deploy email gateways because they don't completely trust incoming email, so we put a security device in the middle.
00:04:42.319 --> 00:04:47.920
That device sees everything, which means we've created an extremely valuable target.
00:04:48.160 --> 00:04:51.600
The same thing is true of a lot of security infrastructure.
00:04:51.839 --> 00:05:11.680
Your firewall sees your network traffic, your EDR platform has enormous access to endpoints, your RMM can execute commands across the environment, your identity provider controls authentication, your SIM contains information about practically everything you're defending, and your secure email gateway sees your email.
00:05:12.000 --> 00:05:21.360
Security products need elevated access in order to do their jobs, but the access that makes them powerful defenders also makes them valuable targets.
00:05:21.680 --> 00:05:25.680
So let's talk about what Fortamil customers should be doing.
00:05:25.920 --> 00:05:28.639
The obvious answer first is patch.
00:05:28.879 --> 00:05:32.240
As of this week, Fortinet has identified fixed builds.
00:05:32.480 --> 00:05:37.279
Fortamil 8.0 customers should move to 8.0.2 or later.
00:05:37.600 --> 00:05:41.839
7.6 customers should move to 7.6.7 or later.
00:05:42.079 --> 00:05:46.240
7.4 customers should move to 7.4.9 or later.
00:05:46.480 --> 00:05:52.879
And systems still running the 7.2 branch need to migrate to a supported fixed release on 7.4 or later.
00:05:53.040 --> 00:06:00.800
If you're listening to this week's later, obviously check Fortinet's current advisory rather than treating those as permanent recommendations.
00:06:01.040 --> 00:06:03.199
But here's the part that I want to emphasize.
00:06:03.439 --> 00:06:07.920
Do not patch the appliance and immediately close the vulnerability ticket.
00:06:08.079 --> 00:06:20.959
If your Fortamil system was exposed while vulnerable, you need to determine whether it was already compromised, because updating Fortamill does not magically delete whatever an attacker may have installed beforehand.
00:06:21.199 --> 00:06:25.759
And Fortinet has actually given defenders some useful things to look for.
00:06:26.000 --> 00:06:31.519
The company published indicators showing attackers modifying and adding files on compromise systems.
00:06:31.759 --> 00:06:35.120
It also documented suspicious configuration activity.
00:06:35.360 --> 00:06:43.920
One example involved creation of an archived account named Archive 234 that was configured to send information to a remote server.
00:06:44.160 --> 00:06:47.360
Think about what that means from a SOC perspective.
00:06:47.600 --> 00:06:56.399
You have an email security appliance, an unexpected archive configuration suddenly appears, and the destination is outside your environment.
00:06:56.639 --> 00:06:59.439
That's not just configuration management anymore.
00:06:59.600 --> 00:07:02.160
That's a potential exfiltration detection.
00:07:02.399 --> 00:07:15.759
Fortinet also documented suspicious cron activity involving a path named/mig admin, modifications to system files, IBE related encryption errors, and unusual administrative events.
00:07:16.000 --> 00:07:21.920
There are also IP addresses and file hashes in Fortinet's advisory that teams can use during triage.
00:07:22.240 --> 00:07:29.920
And I'm not gonna read the SHA 256 hash into a podcast because I would like people to continue listening to this podcast.
00:07:30.079 --> 00:07:35.279
But if you run Fortamail, grab the
advisory and put those indicators into your investigation.
00:07:35.519 --> 00:07:43.519
And there's a second hunting opportunity here that I think is even more important, and that's outbound traffic from security appliances.
00:07:43.759 --> 00:07:56.720
A lot of organizations spend enormous effort inspecting inbound traffic, what's hitting the firewall, what's reaching the VPN, who's scanning the web server, what is somebody sending to our email gateway, which are all good questions.
00:07:56.959 --> 00:08:01.839
But how closely are you watching what the security appliance itself sends outbound?
00:08:02.079 --> 00:08:11.759
Because if an email gateway suddenly starts communicating with an IP address it has never communicated with before, that should be something worth investigating.
00:08:12.000 --> 00:08:17.199
If a firewall suddenly establishes unusual outbound sessions, that's another one.
00:08:17.360 --> 00:08:24.079
If your backup server starts transferring large amounts of information somewhere new, that's a pretty big red flag.
00:08:24.319 --> 00:08:31.279
One of the lessons from this Fort Mail incident is that appliances shouldn't automatically be trusted just because they're appliances.
00:08:31.519 --> 00:08:34.080
They need behavioral baselines too.
00:08:34.320 --> 00:08:36.960
Another lesson here is log retention.
00:08:37.279 --> 00:08:42.960
Bellnet detected the incident in late September, the malicious activity went back to July.
00:08:43.200 --> 00:08:47.440
That's more than two months of history you suddenly need during an investigation.
00:08:47.759 --> 00:08:49.679
Would your organization have it?
00:08:49.919 --> 00:08:53.919
It's easy to say, well, we collect Forty Mail logs.
00:08:54.240 --> 00:08:56.320
Great, but for how long?
00:08:56.639 --> 00:09:00.000
Do you have two months worth or even further back?
00:09:00.240 --> 00:09:03.919
Or does the appliance override everything after seven days?
00:09:04.159 --> 00:09:06.879
Are those logs forwarded somewhere else?
00:09:07.200 --> 00:09:11.360
Can an attacker who compromised the appliance delete the only copy?
00:09:11.759 --> 00:09:15.759
This is why central logging matters and is super critical.
00:09:16.000 --> 00:09:22.240
And preferably security logs should end up somewhere that the system generating them can't simply modify.
00:09:22.399 --> 00:09:30.480
Because once you've discovered an incident, historical telemetry becomes evidence, and you can't retroactively increase retention.
00:09:30.720 --> 00:09:34.080
There's also a broader incident response lesson here.
00:09:34.320 --> 00:09:40.320
So when an actively exploited vulnerability drops, organizations tend to ask, are we vulnerable?
00:09:40.480 --> 00:09:44.240
And that's the first question, but it shouldn't be the last one.
00:09:44.559 --> 00:09:49.440
The next question is, were we vulnerable while attackers were exploiting it?
00:09:49.679 --> 00:09:51.519
And those are different questions.
00:09:51.759 --> 00:10:03.759
Imagine that your Fortamail appliance was vulnerable yesterday, you patch it today, and technically if somebody scans the environment tomorrow, the answer to are we vulnerable is no.
00:10:04.320 --> 00:10:09.279
But the answer to were we compromised could still absolutely be yes.
00:10:09.440 --> 00:10:13.759
And that's the difference between vulnerability management and incident response.
00:10:14.000 --> 00:10:17.679
Vulnerability management asks how do we close the hole?
00:10:17.919 --> 00:10:21.519
Incident response asks, did somebody already walk through it?
00:10:21.759 --> 00:10:30.080
For vulnerabilities that are confirmed to be actively exploited, stocks need to start thinking about both simultaneously.
00:10:30.320 --> 00:10:32.399
And this isn't limited to Fort Mail.
00:10:32.720 --> 00:10:41.279
We've seen this repeatedly with VPN appliances, firewalls, Citrix gateways, exchange servers, so on and so forth.
00:10:41.600 --> 00:10:48.399
A vendor releases an emergency advisory and organizations rush to install the update, which is great.
00:10:48.799 --> 00:10:57.440
But sometimes everybody celebrates after the patch installs successfully while an attacker has been sitting inside the environment for three weeks.
00:10:57.759 --> 00:11:04.000
The patch closes the original door, it doesn't remove the person who already walked inside.
00:11:04.399 --> 00:11:13.120
So I think organizations need a slightly different workflow for vulnerabilities that land in something like Sys known exploited vulnerability catalog.
00:11:13.360 --> 00:11:27.279
Instead of the identify patch, verify close, think of it as identify contain if necessary, preserve evidence, hunt for compromise, patch, and verify, then continue monitoring.
00:11:27.519 --> 00:11:30.960
And the exact order may change depending on the situation.
00:11:31.200 --> 00:11:37.279
Sometimes you have to patch immediately, sometimes you need to capture evidence before rebooting or changing the system.
00:11:37.440 --> 00:11:39.759
And that's an incident response decision.
00:11:39.919 --> 00:11:44.320
But the important part is that somebody actually asks whether compromise occurred.
00:11:44.559 --> 00:11:58.639
There's another very practical exercise I'd recommend after this incident, and that's make a list of your security infrastructure, not your servers generally, specifically the platforms your security program depends on.
00:11:58.879 --> 00:12:13.120
That'll be things like your firewalls, VPN concentrators, secure email gateways, EDR management, SIM, identity systems, RMM platforms, backup infrastructure and vulnerability scanners.
00:12:13.360 --> 00:12:19.279
And then ask, do we monitor those systems as aggressively as we monitor everything they're protecting?
00:12:19.519 --> 00:12:22.480
Because sometimes the answer is surprisingly no.
00:12:22.799 --> 00:12:28.639
Maybe everybody watches firewall alerts, but nobody watches administrative changes
to the firewall.
00:12:28.879 --> 00:12:34.879
Maybe everybody relies on EDR telemetry, but nobody alerts when EDR policies are modified.
00:12:35.120 --> 00:12:43.039
Maybe the SOC reviews phishing alerts from the email gateway all day, but doesn't ingest the email gateway's own administrative audit logs.
00:12:43.200 --> 00:12:44.799
And that's kind of backwards.
00:12:45.039 --> 00:12:51.039
The system enforcing security policy should be more of the most monitored systems in the environment.
00:12:51.279 --> 00:12:57.440
The systems enforcing security policy should be some of the most monitored systems in the environment.
00:12:57.759 --> 00:13:00.320
So here's my challenge for you this week.
00:13:00.559 --> 00:13:06.000
If you use Fortamail, first verify your version and get onto a fixed release.
00:13:06.240 --> 00:13:11.120
Then perform actual compromise assessments using Fortinet's current indicators.
00:13:11.360 --> 00:13:19.600
Look for unexpected archive accounts, look for suspicious CRON activity, and look for the files and hashes Fortinet published.
00:13:19.759 --> 00:13:29.679
Make sure you're reviewing administrative configuration changes, look at outbound connections from the appliance, and make sure you're looking far back enough.
00:13:30.000 --> 00:13:36.080
Based on this bellnet incident, I would not limit the hunt to the day the vulnerability became public.
00:13:36.559 --> 00:13:39.519
Go back months if your telemetry allows it.
00:13:39.759 --> 00:13:43.360
And if you don't use Fort Mill, you still have some homework here.
00:13:43.600 --> 00:13:56.159
Pick one critical security appliance in your environment, maybe your firewall, your VPN, your security gateway, and ask, if this device were compromised today, would our SOC know?
00:13:56.480 --> 00:13:59.519
If not, would the device generate an alert?
00:13:59.840 --> 00:14:00.879
Would you know?
00:14:01.200 --> 00:14:03.519
Do its logs leave the appliance?
00:14:03.759 --> 00:14:06.080
Do you monitor configuration changes?
00:14:06.320 --> 00:14:08.720
Would you detect a new administrator?
00:14:08.960 --> 00:14:11.840
Would you notice file changes on the appliance?
00:14:12.159 --> 00:14:14.559
And how far back could you investigate?
00:14:14.879 --> 00:14:17.759
Because I think that's the real lesson from this story.
00:14:18.080 --> 00:14:21.039
Fortamil exists to inspect malicious email.
00:14:21.360 --> 00:14:25.440
In this case, the security platform itself became part of the attack.
00:14:25.759 --> 00:14:36.799
That's a pretty strong reminder that there's no special category of infrastructure that attackers politely agree not to touch because it has the word security in the product description.
00:14:37.120 --> 00:14:39.919
Security infrastructure is infrastructure.
00:14:40.080 --> 00:14:43.519
It has software, software has vulnerabilities.
00:14:43.759 --> 00:14:49.600
And because we give these systems enormous trust and access, we need to defend them accordingly.
00:14:49.919 --> 00:14:52.960
Patch the vulnerability, but don't stop there.
00:14:53.279 --> 00:14:55.679
Find out whether somebody got there first.
00:14:56.480 --> 00:14:59.440
And that's a wrap for this episode of the SOC Brief.
00:14:59.600 --> 00:15:10.240
If your team is working through the Fortamail vulnerability or you've built something good monitoring around security appliances themselves, hit us up through the website or via social media.
00:15:10.480 --> 00:15:16.879
Keep your appliances patched, keep your logs somewhere safe, keep your eyes open, and keep sharpening those skills.
00:15:17.200 --> 00:15:18.320
We'll talk soon.
00:15:18.559 --> 00:15:19.919
Stay secure out there.
00:15:20.480 --> 00:15:20.960
Bye.