Hello and welcome to the Thursday, August 21st, 2025
edition of the SANS Internet Storm Center's Stormcast. My
name is Johannes Ullrich, recording today from
Baltimore, Maryland. And this episode is brought to you by
the SANS.edu Graduate Certificate Program in
Industrial Control System Security. In diaries today, I
wrote a little bit about some of the odd usernames and
passwords that we often see in our SSH and Telnet data that
doesn't necessarily link to a direct attack, but maybe some
of you have ideas what's behind that. The username that
prompted this was Airtel@123. Quick, Google shows that,
well, this does link to a company called Airtel that
disputes a router. However, this password is not the SH
and Telnet password. Way too complex for that. That's
actually just admin admin for this particular router. But
this password is used for the Wi-Fi network by default. So
that's kind of odd why they're using that. I don't know.
Maybe some users just get lazy and when they have to change
the admin password, they're just changing it to the WPA
passphrase. That could be a motivation behind it. Also,
another username-password combo that I thought was kind
of interesting, not root admin, but instead they
replaced the first letter with the dollar simple. So dollar
OOT and then dollar D-M-I-N. Again, if anybody has any idea
why this may happen, whether by accident or intentionally,
please let me know. Apple today fixed a single
vulnerability in iOS, iPadOS and macOS, going back to
versions for macOS. The reason for this quick patch for one
individual vulnerability is that this vulnerability in
Image.io is already being exploited. It's a memory
corruption vulnerability leading with that to arbitrary
code execution. So definitely a patch that you do want to
apply quickly, even though at this point, of course, it has
only been cited against, well, very targeted attack victims.
And then we got an interesting issue with Microsoft Copilot
that was nicely documented by Zach Korman. And with that
also a little bit controversy how Microsoft dealt with this
particular problem. The core issue here are audit logs. You
expect your audit logs to record if a particular user is
accessing some specific information. Well, this is not
the case if that information came from Copilot. Copilot
indexes basically various files on your system. So
basically Copilot accesses those files. A user can now
ask questions about these files. And essentially the
Copilot system is returning very specific answers about
the content of these files. But there is no access log
that will actually record that a particular user did retrieve
that data. And that's sort of a little bit the problem of
this issue and the controversy around it that Microsoft
didn't necessarily acknowledge this as a bug properly. But
just from the technical point of view here, this is
certainly something to be aware of. This has been some
ongoing issue where really access control is sort of
getting destroyed by AI agents. Any data that the AI
agent is being trained on is accessible to users that have
access to the AI agent. And as a result, well, any sort of
more fine-grained access control that you may have had
on a per-user basis on the original data is kind of lost.
And with that, of course, also the user's perspective audit
capabilities.
And Marek Toth, who presented about this topic at DEF CON,
did update a blog post about click-jacking vulnerabilities
in password managers. The problem with click-jacking
password managers arises from password managers essentially
inserting themselves into web pages. And in doing so,
they're susceptible to many of the attacks that web
applications and HTML and the DOM are susceptible to,
including click-jacking. And turns out that click-jacking
is, well, quite widespread among the different password
managers. Marek tested a number of different password
managers, pretty much all the well-known ones. Sadly, many
of them have not been fixed at this point. NordPass,
ProtonPass, RoboForm, Dashlane, Keeper, those are
the ones that are fixed. Bitwarden, 1Password, iCloud
passwords, nPass, LastPass, LogMe ones are not yet fixed
and are vulnerable to some extent. Now, for some of them,
for example, 1Password, credit card numbers are not affected,
but passwords are. And that's probably almost more important
than credit card numbers. So watch out for updates from
various password managers and, well, apply them because as so
often with click-jacking, the exploit is actually pretty
straightforward. Well, and that's it for today. So thanks
again for listening. Thanks for subscribing, liking,
commenting, and recommending this podcast and talk to you
again tomorrow. Bye.
Bye.