Defensive Security Podcast - Malware, Hacking, Cyber Security & Infosec
Sep 29, 2026 · 1 hr 6 min · 9 segments
Please consider supporting the DefSec podcast here. 1\. AI agents broke into 395 organisations, including ones their operator ruled out…
Jerry BellHost
Andrew KellettHostFirst up,
we
I thought this one was good to talk about because it gives a little bit of insight into how some of the adversaries are actually kind of weaponizing the AI, not, you know, it's, I think for a lot of people, it's kind of this dread risk.
And other than the, you know, the, the big fiasco with open AI and hugging face, there's not been a ton of like actual reporting on how this is actually playing out.
And by the way, I'm not sure which one it is because there's been so many gosh darn paper cut vulnerabilities over the past couple of weeks.
with Papercut and an Active Directory server and used AI to basically poke at it and develop an exploit chain and turned it loose.
But they had apparently put some constraints on it that it should not attack targets in certain countries however the the company who or the security company who found this i think it was gray noise basically found that you know it was attacking everybody indiscriminately it was going after victims kind of regardless of of country which is not too surprising i don't I don't think a lot of people have a great handle on how to use these tools yet.
I don't know if they're using Metasploit or whatnot through like an MCP server, but they created some kind of custom chain to go after their targets, and it was, well, quite effective.
The reporting says that 11 of the organizations were compromised in like 26 seconds.
And there was a high school that was among the victims that had their domain admin account compromised in seven minutes.
And that the whole shebang took like four hours to operate, which is kind of scary.
And I think really highlights the challenge that lies ahead of us, which is, you know,
A,
right?
First up,
we
I thought this one was good to talk about because it gives a little bit of insight into how some of the adversaries are actually kind of weaponizing the AI, not, you know, it's, I think for a lot of people, it's kind of this dread risk.
And other than the, you know, the, the big fiasco with open AI and hugging face, there's not been a ton of like actual reporting on how this is actually playing out.
And by the way, I'm not sure which one it is because there's been so many gosh darn paper cut vulnerabilities over the past couple of weeks.
with Papercut and an Active Directory server and used AI to basically poke at it and develop an exploit chain and turned it loose.
But they had apparently put some constraints on it that it should not attack targets in certain countries however the the company who or the security company who found this i think it was gray noise basically found that you know it was attacking everybody indiscriminately it was going after victims kind of regardless of of country which is not too surprising i don't I don't think a lot of people have a great handle on how to use these tools yet.
I don't know if they're using Metasploit or whatnot through like an MCP server, but they created some kind of custom chain to go after their targets, and it was, well, quite effective.
The reporting says that 11 of the organizations were compromised in like 26 seconds.
And there was a high school that was among the victims that had their domain admin account compromised in seven minutes.
And that the whole shebang took like four hours to operate, which is kind of scary.
And I think really highlights the challenge that lies ahead of us, which is, you know,
A,
right?
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.