security researcher/programmer/director @ HexArcana Cybersecurity GmbH ⁂ @pagedout_zine ⁂ @DragonSectorCTF ⁂ https://nitter.cf/t.co/ShG2c5As1K ⁂ ex-Google ⁂ he/him
Zürich, Switzerland
Joined July 2009
- Tweets7K
- Following1.3K
- Followers39.1K
- Likes11.7K
Gynvael Coldwind retweeted
To expand on this a bit: Yes, the LLM found more bugs than the fuzzer. But the fuzzer finding bugs the LLM didn't shouldn't come as a surprise to anyone who has ever performed a thorough manual review, only to run a fuzzer afterward and discover issues they missed.
People often ask me how I think about fuzzers and other tooling in the age of AI. Instead of my opinion, here is some hard data by @5aelo.
Gynvael Coldwind retweeted
Universal exploit for #php2shell is rather impossible, but specialized exploits for known environments are coming. I tried my luck with pfSense 2.7.2 - PHP 8.2.11, FPM, 512M limit. Variant with session file write to a crafted path. Almost entirely written by an open-weights LLM.
Gynvael Coldwind retweeted
Quake for Atari 8-bit now available!
kweepa.itch.io/quake64
With all these exploits flying around on the internet I think we should all switch to HTTP 4.0, which I would like to propose here:
HTTP 4.0: Human hand-To-hand Transfer Protocol
A post-apo protocol where you just hand your friend your website on a pendrive/CD/floppy/tape.
You MUST first check if they are not a replicant of course.
There are a technical lot of things I will miss that AI basically made obsolete.
But one thing I won't miss is DEBUGGING RANDOM ISSUES ON MY LINUX DESKTOP! AI deals with them super fast - basically a solved problem. And good riddance.
Year of Linux on desktop?
Usual reminder that Paged Out! magazine now has Patreon, so if you'd like to support the zine and help to fund next issues, please start here: patreon.com/PagedOut
Gynvael Coldwind retweeted
float i,e,R,s,k,z;vec3 q,p,d=.5-FC.rgb/r.y;for(q--;i++<1e2;i>85.?d/=-d,o+=(cos(vec4(1,3,5,0)+k)-k)*e*.1:o){s=3.;q+=d*e*R*.2-4e-5;p=vec3(log(R=length(q))-t,acos(q.z/R)-.5,atan(q.x,q.y)+t*.5);for(e=--p.y;s<4e3;s*=3.)e+=(z=sin(dot(p/p,exp(cos(p*s)+1.))))/s,k+=z/70.;}
#つぶやきGLSL
Want to learn K8s hacking to be able to defend your or your client's infra? We're running an amazing hands-on workshop starting next week!
Sign up here:
hackarcana.com/workshop-sess…
Gynvael Coldwind retweeted
From Sunday's Ultima VII Revisited stream: prototype "Real-Time With Pause" combat for Ultima VII!
[PL] Nagranie z mojej prelekcji o (pseudo)terminalach pojawiło się na YouTube :)
youtube.com/watch?v=Znbc4MDY…
Shower thought: ArtCraft (that clean room Adobe suite reimplementation) is released on MIT and Apache licenses.
But that sounds incorrect. It's said to be purely vibe coded, which makes it ineligible for any copyright protection, right? So it should be public domain?
Gynvael Coldwind retweeted
I’m seeing a lot of Slibes (slop-vibe-slides) lately. AI can throw together a pretty-looking slide deck faster than you can blink. However, just like slop-text, slop-slides show a lack of respect for your audience. If it’s worth my time to read, it’s worth your time to edit.
Gynvael Coldwind retweeted
Agree on verification, but I'd push it further. Memory safety is a symptom, not the cause. We got here because of design decisions and trade-offs we made all along, and you can see it in pretty much every recent Linux kernel disclosure.
Throwing more tokens at the attack surface won't save us either. Nothing we have today is ready for AI-driven vuln research at this scale, and finding and fixing bugs with AI doesn't change that.
What we actually need is redesign and architecture validation with AI, formal methods where we can, and not accepting attack surface at the design phase in the first place. That's just not how we build software today. Patching known bugs doesn't touch the root cause, and hardware has the same problem.
Security will become a larger and larger function in software companies. Security is verification engineering (eg: “my code is probably memory-safe”), as well as capital allocation (“what surface should I throw most tokens at?”)
For startups, it’s a challenge and opportunity. As the world’s paranoia grows given ever-more-sophisticated AI adversaries, how can you trust a 2-person-and-a-dog company?
But conversely, given the dire state of global cybersecurity and increasing reliance on centralization, as always, there’re vast areas that small teams can now disrupt that I’m excited to invest in.
Reminder: Paged Out! #10 call for papers deadline is TODAY end of day (anywhere-on-earth timezone)!
If you're in or around Zurich, I'll be speaking about bit-perfect recompilation at vulnz event next week ↓
vulnz.ch v0.7 will take place on Wednesday, October 7th, and will feature two talks.
⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯
𝐑𝐞𝐜𝐨𝐯𝐞𝐫𝐢𝐧𝐠 𝐒𝐨𝐮𝐫𝐜𝐞 𝐂𝐨𝐝𝐞: 𝐁𝐢𝐭-𝐏𝐞𝐫𝐟𝐞𝐜𝐭 𝐑𝐞𝐜𝐨𝐦𝐩𝐢𝐥𝐚𝐭𝐢𝐨𝐧 𝐔𝐬𝐢𝐧𝐠 𝐂𝐥𝐞𝐯𝐞𝐫 𝐓𝐫𝐢𝐜𝐤𝐬 𝐚𝐧𝐝 𝐀𝐈 (@gynvael)
Getting a decompiled project to compile is one thing. But what if you wanted to recompile a piece of software into a byte-for-byte identical executable? This talk is a case study of achieving a bit-perfect recompilation of an old DOS game, from the initial idea and manual reverse-engineering work to the design of a fairly complex, regression-proof system that ensured the AI didn’t meander while decompiling and lifting the code.
𝐀𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐧𝐠 𝐀𝐩𝐩𝐒𝐞𝐜 𝐰𝐢𝐭𝐡 𝐀𝐠𝐞𝐧𝐭𝐬 (𝐑𝐲𝐚𝐧 𝐒𝐞𝐚𝐫𝐥𝐞)
We’re transforming traditional security reviews with agents. Learn how we’re gathering application context, generating threat models, and identifying and fixing control gaps and vulnerabilities.
⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯
@snyksec, the AI-driven developer security platform, has an office at Zollstrasse 17, next to Zurich’s main railway station. Snyk has kindly offered vulnz.ch a new home, and its office will serve as the meetup’s new venue.
If you haven’t joined us at vulnz.ch before and you’re a Zurich-based professional working in appsec, pentesting, vulnerability research, AI security, or anything in between, come along!
luma.com/v0etnoj3
