
@pksecurity_io | KAIST Hacking Lab | Microsoft MVR
Joined February 2022
- Tweets161
- Following90
- Followers604
- Likes658
This research was just a personal project for fun, but the work being done at @pksecurity_io is far more impressive and on a completely different level.
After the advisory is released by SSD, I'll be sharing some of the more interesting details on the @pksecurity_io blog.
Our researcher, @justlikebono, successfully demonstrated a kernel exploit in the Windows LPE category at TyphoonPwn 2026.
We will share the technical details on our blog after the vulnerability is patched.
We are continuously conducting security research on Android, Windows, and IoT. If you’re interested in our work, feel free to reach out to us anytime!
zdi probably got flooded with slop-software bugs, local inference stacks, nvidia, ai databases, and coding agents.
they should’ve filtered a lot of these submissions tbh. for half of this stuff, you just need a minute with claude or codex to get shell.
they should have accepted submissions on browsers, virtualization, lpe, server-side targets, and containers
‼️🚨 Pwn2Own Berlin 2026 just hit a wall. For the first time in 19-years, ZDI rejected dozens of working zero-day RCE submissions because organizers ran out of contest slots.
Rejected hackers are now going public with PoC demos and direct vendor disclosures, breaking Pwn2Own's usual secrecy.
▪️ AI surfaces a massive wave of 0-day RCEs.
▪️ Submissions overwhelm ZDI past max capacity.
▪️ Slots run out. Researchers with working chains get rejected.
▪️ "Revenge disclosures" begin. ← we are here.
Confirmed casualties so far:
▪️ @xchglabs : 86 vulnerabilities prepared (PyTorch, NVIDIA, Linux KVM, Oracle, Docker, Ollama, Chroma, LiteLLM, llama.cpp). All rejected. Now reporting directly to vendors with writeups dropping as patches land.
▪️ @ggwhyp : full-chain Firefox RCE on Windows. Rejected. Publicly demoed (HTML page → cmd.exe → calc.exe). Responsibly disclosed to Mozilla.
▪️ @yunsu_dev : working RCE chain, rejected. Submitting elsewhere.
▪️ @ryotkak : tried to register for 3+ weeks. ZDI confirmed "at maximum capacity, can't add extra contest days." Considered canceling flight and hotel.
▪️ @anzuukino2802 : Claude Code RCE PoC. Rejected.
▪️ @desckimh : 0-day RCEs in Ollama and LM Studio. Rejected.
Reported impact: a community-estimated 150+ researchers tried to register. Accepted contestants are now being warned about collisions. Rejected vulnerabilities going to bug bounty programs may trigger pre-event patches that invalidate the work of those who got in.
ZDI has not publicly addressed the capacity issue. The event still runs May 14-16 in Berlin.
MSRC is really ridiculous. I don't know if they're refusing bounties or what. I accidentally discovered and submitted OOB Write and UAF vulnerabilities, and they either said the vulnerabilities were low-risk or that they couldn't reproduce them in the latest version.
Kim Dong-uk retweeted
Finding a bug isn't as mentally challenging as researching what exactly you can overflow/replace/confuse it with. A bug is a known unknown. Overwrite target is unknown unknown in a lot of cases, if your research isn't a continuation of somebody else's.
Kim Dong-uk retweeted
Whoah... $250000
(CVE-2025-4609, similar to CVE-2025-2783/412578726)[412578726][Mojo][IpczDriver]ipcz bug -> renderer duplicate browser process handle -> escape sbx is now open with PoC & exploit(success rate is nearly 70%-80%)
issues.chromium.org/issues/4…
issues.chromium.org/issues/4…
(CVE-2025-4609)[412578726][Mojo][IpczDriver]Incorrect handle provided in unspecified circumstances
chromium-review.googlesource…
Reported by Micky on 2025-04-22
Kim Dong-uk retweeted
Little late, but yea.
following last year's USACyKor, this year's CyGoN get second place in codegate 😀
Kim Dong-uk retweeted
BYOVD is no longer required on modern Windows. a bunch of recent msft kernel patches just shove vulnerable kernel code behind a privilege check. I wonder if we will see threat actors leveraging this like they did BYOVD. no bringing along a driver and loading also means less IOCs
Kim Dong-uk retweeted
(CVE-2025-2783)[405143032][mojo] Google Chrome Sandbox Escape(exploited ITW) is now open(PoC & exploit are restricted🥲)
issues.chromium.org/issues/4…
Reported by @oct0xor @2igosha
Exploited ITW
(CVE-2025-2783)[405143032][mojo]
Receiving/sending sentinel handle values over IPCZ -> misinterpreted by OS function(Windows) -> ..
chromium-review.googlesource…
learn.microsoft.com/en-us/wi…
issues.chromium.org/issues/4…
chromereleases.googleblog.co…
@oct0xor @2igosha
Kim Dong-uk retweeted
When working on certain hard exploits, you might dread & fold from how difficult the task is; true, some exploits are nearly PhD-level problems, verily so. But there will be no gain without this pain. You either go through it or go away. There is no royal path.
Nice work.
I've previously delved into this, and since Windows Defender is closely integrated with Windows OS internals, effectively fuzzing it requires significant engineering effort.
But it's definitely possible. :)
Fuzzing Windows Defender in the Honggfuzz+IntelPT (hardware) mode
scrapco.de/blog/fuzzing-wind…
by @buherator
4
1,057
Kim Dong-uk retweeted
mozilla.org/en-US/security/a…
Reported by Dong-uk Kim (@justlikebono)
Kim Dong-uk retweeted
Over the past few days, the IDA Pro (reverse engineering) MCP server—let’s call it Re MCP—has become a hot topic. In my opinion, it’s not yet capable of automatically uncovering vulnerabilities. However, if we look at it from a different angle, Re MCP is akin to tools like CodeQL or Hex-Rays in that it significantly reduces the time and effort required for analysis.
Hex-Rays was revolutionary in the era of pure disassemblers by introducing a decompiler that dramatically cut down the costs of reverse engineering. Similarly, CodeQL freed analysts from the tedious process of reading through code line by line to identify bug classes.
Likewise, Re MCP improves the readability of decompiled code by renaming functions and variables, summarizing logic flows, and inferring data structures. These features substantially lessen the reverse engineering overhead for vulnerability analysts, allowing them to focus on actually finding security flaws.
In short, I’m definitely a fan of this tool.
Kim Dong-uk retweeted
Everyone wants to fuzz complex optimizations of WebAssembly compilers, but most approaches are simply considering grammars and types. Why can't we fuzz optimizations directly? We will share how we can do if we have systematically written optimization rules in S&P 2025 :)
Kim Dong-uk retweeted
Since I'm 6 drinks in for 20 bucks, let me tell you all about the story of how the first Microsoft Office 2007 vulnerability was discovered, or how it wasn't.
This was a story I was gonna save for a book but fuck it, I ain't gonna write it anyways.
Kim Dong-uk retweeted
I just pushed out wtf v0.5.5 where you can now generate full execution traces with both hypervisor backends (winhv & KVM) 🔥🔥
This should hopefully greatly help people debug their setup 🪄
github.com/0vercl0k/wtf/rele…
Kim Dong-uk retweeted
We HypeBoy topped Plaid CTF 2024! 👑
@_Reinose_en @justlikebono @munsiwoo @real_as3617 @mhibio_ptw @howdays1 @stanzz @soo_sudo @n1net4il @1nteger_c @c2w2m2 @pr0cf51 @cmy981224 @csapp3e ironore15 G0RiYa
Thanks to the great support from dakuo & hellsonic





























