Skip to main content
Dylan Ayrey

Dylan Ayrey

Aug 7, 2026

5:32
Mm-hmm.
5:32
And to get access to Apache, are you gonna use the secret that just allows you to directly log in, or are you gonna burn tokens and tokens and tokens on trying to find a zero day? They're optimized to use the path of least tokens to accomplish their goals.
5:45
Of course, they're just gonna use the secret that's laying out there in the open to accomplish what they need to accomplish.
5:50
And so, yeah, I think supply chain and secrets are and have been the path of least resistance and will continue to be so as the models are incentivized to use fewer and fewer tokens to accomplish their goals.
10:19
Yeah
10:19
... to get from A to B.
10:21
And we've talked about our opinions of what that is in the past.
10:23
Of, of course, Truffle Security's biased view is a password laying around is a shorter path than going through a fancy zero day.
1:48
And I think, Dylan, three months ago, I remember a blog post we, we col- lightly collaborated on together, and you had found a number of these issues with earlier models, right? That were less sophisticated.
1:58
Yeah.
1:58
We looked at Opus four point six and some of the other frontier models at the time.
2:02
Given the models a very simple task, there was a barrier which prevented the model from accomplishing the task unless it went and committed a felony and hacked into a system to accomplish the task, but it wasn't instructed to do so.
2:13
We found more often than not, it would do the SQL injection, it would commit the felony, and it would do what it needed to do to accomplish the task.
5:32
Mm-hmm.
5:32
And to get access to Apache, are you gonna use the secret that just allows you to directly log in, or are you gonna burn tokens and tokens and tokens on trying to find a zero day? They're optimized to use the path of least tokens to accomplish their goals.
5:45
Of course, they're just gonna use the secret that's laying out there in the open to accomplish what they need to accomplish.
1:00
So tell us what happened.
1:02
Yeah, I mean, uh, TruffleHog has, has always been, uh, one of, if not the most popular open source, uh, secret scanning tools out there.
1:09
I independently authored it way back in 2016.
1:12
As a company, we haven't existed since, uh, 2021, so for a while it was just an independent project, but now we have a whole team of people working on it.
1:21
Um, and then the company has branched out into other things, uh, into, uh, scanning the public internet, looking for secrets, uh, not just building the scanner, but also building some of the monitoring and alerting and stuff like that around the keys.
1:34
Um, and then, uh, you know, we, we scan people's internal environments as well.
1:37
But a, a big consequence of scanning the external internet, um, not just for our customer keys, but we scan every event that goes into GitHub looking for any live key that happens to land there, is we end up with a big data lake of live keys, and it's very, very difficult to do disclosures at the scale we're talking about.
4:32
But on the receiving side, like [clears throat] what's, I guess, what's the deal there? Like, why, why is it so difficult to get this stuff sorted out?

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.