My mom basically told me the she would like to be a senior UAT tester - as in telling clueless kids how old people can't figure out their UI.

We should definitely have that role!

A fool with a tool is a more dangerous fool.

— u

The Google Project Zero post from a while ago that showed memory safety bugs were more common in new code continues to annoy me because it’s such a clear example of computer scientists doing social science without understanding how to do it.

The root problem is that they are no counting the number of bugs. There is no ground truth that they are able to use. They are counting the number of bugs found. And how are bugs found? When you have some existing technique (static analysis, fuzzing, code review, whatever) that finds that kind of bug.

So their conclusion is ‘bugs that existing techniques are able to find are less likely to appear in code that has already been subject to those techniques’. To which the correct response is ‘well, duh’.

The original Coverity paper, which introduced a large range of brand new static analysis techniques, found a load of bugs in old code. When we started running existing C and C++ code on CHERI, we saw very little correlation between the bugs that we found and the age of the codebase (we found quite a few bugs that were 20+ years old).

If you remove the sample bias, the results are far less clear cut and there are a load of confounding factors. If a memory-safety bug causes random crashing in one in a million runs of a userspace program based on some non-deterministic factor, it’s unlikely that anyone will debug it. If a sequence of actions reliably crashes a kernel, it will probably be fixed. If an attacker finds it and is actively exploiting it, it will almost certainly be fixed. The symptoms and reproducibility of the bug have a huge impact on whether it will be fixed.

Coolest AI project I've seen so far:

https://www.youtube.com/watch?v=IAuapNwJ2vQ

Praise be the Omnissiah!

#wh40k

Hashcat v7.0.0 released https://hashcat.net/forum/thread-13330-post-63567.html#pid63567

Proton’s Lumo AI chatbot: not end-to-end encrypted, not open source

https://pivot-to-ai.com/2025/08/02/protons-lumo-ai-chatbot-not-end-to-end-encrypted-not-open-source/ - text
https://pivottoai.libsyn.com/20250802-protons-lumo-ai-chatbot-not-end-to-end-not-open-source - podcast
https://www.youtube.com/watch?v=HDPZbUPUFyk&list=UU9rJrMVgcXTfa8xuMnbhAEA - video

#Linux eBPF vulnerabilities incoming (unprivileged eBPF required) + disclosure troubles:

https://www.openwall.com/lists/oss-security/2025/08/03/1

@freddy @ifsecure Here's the official announcement, FTR: https://www.mail-archive.com/security-announce@lists.apple.com/msg00842.html (APPLE-SA-07-30-2025-1 Safari 18.6)

Based on the previously linked issue it looks like the patch window was this big due to the misalignment of patch cycles, no?

a 2661 byte program I wrote just won the "Sur Prize" at the International Obfuscated C Code Competition. You can probably guess what it is once I mention that @foone might enjoy it

https://www.youtube.com/watch?v=d2ulsnSTbUQ

[RSS] Exploring possible solutions to the inconsistency in how Windows searches case-insensitively for named resources

https://devblogs.microsoft.com/oldnewthing/20250723-00/?p=111403

Some fun anti-reverse possibilities here :)

How to craft a raw TCP socket without Winsock? https://leftarcode.com/posts/afd-reverse-engineering-part1/

From a CBS news segment from July of 1985 discussing the busting of various #hackers and BBS operators in New Jersey.

Ouch, but also 💀

Will be uploading the entire segment to Internet Archive later today.

Another day, another conversation with the press team where I explain that I did not give the quote in that story and the whole thing is AI slop. This happens once every few weeks now.

Why does [ #WinDbg ] show me the wrong function?

https://devblogs.microsoft.com/oldnewthing/20050322-00/?p=36113

TIL about COMDAT folding #compiler optimization!

[RSS] Exploit development for vulnerabilities in Windows over MS-RPC

https://incendium.rocks/posts/Exploit-Development-For-MSRPC/

New vulnerability report from Talos:

Eclipse ThreadX FileX RAM disk driver buffer overflow vulnerability

https://talosintelligence.com/vulnerability_reports/TALOS-2024-2088

[RSS] MaterialX and OpenEXR Security Audit

https://www.shielder.com/blog/2025/07/materialx-and-openexr-security-audit/

[RSS] Characterizing the Raspberry Pico 2 FI countermeasures - Part 1

https://www.ioactive.com/characterizing-the-raspberry-pico-2-fi-countermeasures-part-1/

In 1983, Philips produced the first FM radio receiver on a chip, leading to products such as the FM radio wristwatch. Let's look at the tiny silicon die inside this chip and see how it works. 1/N