@gdb Formal verification is becoming much more practical, as are complete reimplementations of C programs in memory-safe languages!
I have some agents chewing through a firmware rewrite in rust with no input from me. It is a new eraβ¦
Fixes are available for CVE-2026-109518, a High severity vulnerability in APT, the package manager used in Debian, Ubuntu, and derivative Linux distributions. The CVE has a CVSS 4.0 score of 7.7.
Remediation details are available on our blog.
https://t.co/x3q5jIeD4J
I used an AI for two hours and found a zero-day that would let a nation-state hack any server running stock Ubuntu. It was patched this morning. CVE-2026-109518. π§΅
π¨ BREAKING: A flaw in APT, the tool that installs software on Debian Linux, lets an attacker on the network tamper with a package after it passes verification and run code as root.
The bug, CVE-2026-109518, sits in resumed HTTP downloads. APT can merge two different downloads when a redirect gives them the same address, and a man-in-the-middle can then alter the file after its check.
Debian fixed it in apt 3.0.4, shipped to Debian 13 security updates on 4 November. Users should upgrade now.
Per Debian's changelog, the fix is four words long. "Open target files exclusively."
Anyway, I don't work professionally in security anymore, and I found both of these bugs in a couple days. I can only imagine what one could do with more time and access to less neutered models. We should expect some extremely crazy bugs to be found and exploited over the next few months. This might be obvious to most folks here, but there is currently huge alpha in taking some code or binary someone has not audited yet and just aiming a powerful AI model at it.
@matthew_d_green It's OK but in my tests GPT 5.6 or GPT 6 Sol with daybreak blue enabled is way better for these tasks. Pretty easy to get access, too.
@haoxingdu@dwarkesh_sp I have a chatbot hooked up to telegram that I use for my journal/daily tasks, and it dropped down to opus 4.8 when I mentioned reporting some security bugs in software :(
@haoxingdu It's funny because you can't even give a *hint* of a preference, or these models like to run with it. You can't say "option A makes sense to me for these reasons, but honestly it seems like a wash, do whatever you think is best", because it'll go with option A like every time.