🤙💰 Mahalo FIN7: fireeye.com/blog/threat-re…
• On several on-going investigations we saw #FIN7 trying to retool 🏄🏼
• Used DLL search order hijacking of a legit POS management utility with a signed backdoor (0 detections on VirusTotal)
• Hunting for #BOOSTWRITE and #RDFSNIFFER 💳
.@josh__yoder & I stayed up much of the night to get this blog out.
The signed #BOOSTWRITE sample is still undetected by static VT scanners: virustotal.com/gui/file/18cc5…
We were fair on why that is and how that doesn't fully represent detection posture.
Then we provided hunting rules.
#FIN7's code signing certificate is purportedly from Mango Enterprise Limited in the UK.
Prob not theirs - based on the street address, I suspect there's more car theft than certificate theft 😜: maps.app.goo.gl/MbznDeJPHJr4n5…
We analyze & discuss how to find the certificate anomalies!
Btw, here's an awesome Yara certificate anomaly signature concept from @wxs in a thread discussing them, including @NCCGroupInfosec's rule from @edeca that's featured in today's #FIN7 blog:
Oh and Google/@virustotal crew was great to work with on the bug!
We tried to share a few examples of #AdvancedPractices 🦅 leveraging certificate, PDB path, and export features for discovery.
I snuck some content in those rules 🌶️ but they are inspired by @stvemillertime, who I've decided has a PhD in hunting
↘️fireeye.com/blog/threat-re…
In a perfect confluence *cough* of events, @BarryV talked about #FIN7's historical code signing of binaries, scripts, and even documents @ #FireEyeSummit:
Note the operator environment details we extract for FIN7, like their LNKs
CC @keydet89, @silascutler
Hey so, we accidentally included a #Turla REDUCTOR export DLL name in one of the blog's Yara rules. 🤦♂️ Whoops!
Turla ≠ #FIN7
🧢 h/t @OleVilladsen for the catch!
The value of our (+@cglyer) real-time attacker technique collaboration with absolute beasts in the industry @doughsec 😶🌫️, @penninajx + @srunnels 💻 cannot be overstated bringing together puzzle pieces for the RE wizards on each side
From the new @WIRED article: wired.com/story/the-unto…
———
While Mandia conferred with the government, Charles Carmakal, the CTO of Mandiant Consulting, contacted some old friends. Many of the hackers’ tactics were unfamiliar, and he wanted to see whether two former Mandiant… twitter.com/i/web/status/1…
We've been tracking DEV-0537 since 2021 (overlaps: Lapsus$, UNC3661). Here's a comprehensive 🆕 BLOG 📰 covering observed TTPs: microsoft.com/security/blog/…
#MSTIC and Defender threat intel collab
➕#DART 👻 incident response team experience from the trenches [1/3]
The blog highlights varied initial access vectors and a slew of [inconsistent?] end goals: data theft, extortion, chaos...
One way to interpret "this actor's TTPs and infrastructure are constantly changing" is that they are loosely-organized (see:
DEV-0537 / Lapsus$ shows that 𝘢𝘵𝘵𝘢𝘤𝘬𝘦𝘳𝘴 can be creative opportunists and still be successful.
Luckily, the same goes for 𝘥𝘦𝘧𝘦𝘯𝘥𝘦𝘳𝘴.
Use this opportunity to strengthen your security controls to protect far beyond this threat actor [3/3] microsoft.com/security/blog/…
But then what?? Let’s talk about some post-compromise techniques...
Please read the above blog to appreciate multiple backdoors used, careful & unique tradecraft used on-premise...
We just published more details on what we’ve been finding post-compromise: blogs.microsoft.com/on-the-issues/…
ADFS key material compromise, SAML shenanigans, OAuth keys added...
Yes - it's hardcoded for netblocks released in the #MSTIC report (microsoft.com/security/blog/…)
This is just extra coverage on top of existing cred harvesting logic
That said, the logic posted there finds some high fidelity #STRONTIUM campaigns from at least June through... recently (more details in above blog).
You'll see a User-Agent, first/last attempt, # of total attempts, # of unique IPs & unique accounts attempted + a list of accounts
As shipped, it's looking over the past 30 days. But if you have #AzureSentinel, I recommend pasting that same KQL in & searchings logs w/ expanded timeframe.
The # authAttempts can stay where it's at ... #STRONTIUM activity is approx 100 attempts per IP per account