Pinned Tweet
#Qualcomm 4/5G #Baseband internal messaging system and state machine reverse engineering , github.com/vessial/baseband/… ,only Chinese version
New Blog post:
ROOT Tesla OS on QEMU Part 1 :unpacking cn0xroot.wordpress.com/2026/…
NVIDIA GPU backend falcon controller IDA pro disassembly plugin github.com/vessial/nvidia
open source the TI DSP chip TMS320F2837xd series disassembly IDA Pro plugin github.com/vessial/ti
GLM-5.2 is Fully Open, Frontier Intelligence Belongs to Everyone
Today, the sudden restriction of certain frontier models is deeply regrettable. At a time when access to frontier models is abruptly cut off for non-technical reasons, we are even more convinced of one thing: science should be global.
The path to AGI (Artificial General Intelligence) must never be enclosed by high walls. We have always believed that AGI should be the cornerstone for all of humanity to collaboratively explore the boundaries of intelligence and solve complex challenges, rather than a privilege monopolized by a few rules and subject to revocation at any moment. In the face of external blockades and restrictions, our attitude is one of radical openness. Frontier intelligence must remain open-source, accessible, and buildable, serving every dedicated developer.
GLM-5.2 is Zhipu's most capable open-source model to date. It not only supports a truly usable 1M context window but also maintains a continuous lead in the independent completion of long-horizon tasks, providing solid foundational support for building complex agent applications. It also continues to be our main engine for creating the strongest domestic coding model.
Tonight at 5:21—at this special moment—GLM-5.2 will officially be available to all GLM Coding Plan users (including Lite / Pro / Max). The API will also go live next week.
A step closer to frontier intelligence for everyone.
The future of AI is open, and it is for the people.
ModelKey: GLM-5.2
EXPMON has been updated to v20260202! This is a Detection Logic update, to provide decisive Detection Result against the ongoing Microsoft Office zero-day exploits CVE-2026-21509.
For example, below are two real, in-the-wild samples of the Office 0day which were just disclosed today:
pub.expmon.com/analysis/3116…
pub.expmon.com/analysis/3116…
Previously, these samples were detected only via the Indicators (see xitter.cf/HaifeiLi/status/201835…). Since this update, they will now be classified/shown as "Malicious - potential exploit CVE-2026-21509", easier for users to identify the threat.
Enjoy the hunting!:)
#expmon #CVE-2026-21509 #office #zeroday #0day #threatintel
Someone submitted the real CVE-2026-21509 sample to EXPMON last night!
Check out this submission:
pub.expmon.com/analysis/3116…
The SHA256 of the sample, it's a RTF file:
c91183175ce77360006f964841eb4048cf37cb82103f2573e262927be4c7607f
While EXPMON didn't report the 0day immediately - this is well expected, it reported various highly suspicious Indicators, including the key Indicator named "activex compatibility shellexplorer registry key accessed". I shared how to use this key Indicator on EXPMON to hunt the 0day just days ago: xitter.cf/HaifeiLi/status/201634….
That's enough to investigate it manually in your local env, and I have just confirmed this is indeed the CVE-2026-21509 zero-day exploit!
My quick analysis showed that this is the initial attack vector sample in a full attacking chain. Thanks to EXPMON logs, I quickly found that the RTF file was trying to load the IE engine (the "ieframe.dll") while also trying to connect to the threat actor controlled server, one of the url is "\\wellnesscaremed[.]com\davwwwroot\venezia\Favorites\blank.doc" (see the pic attached) - all activities are automatic, meaning as soon as the victim open the Word document, the victim could be pwned. There're not just one but quite serveral OLE objects in the RTF which are, to be honest, quite sophisticated, showing the sophistication of the zero-day attack. Full details haven't been fully understood in such a short time.
The same sample was also listed/confirmed by the Ukrainian CERT cert.gov.ua/article/6287250 independently, there're more details in that article please go check out.
What a wild story! Thank you very much to the person who submitted the sample (and I received your message about adding the "unzip" feature:))! Once again it confirmed the effectiveness of the EXPMON system when it comes to detecting unknown 0day exploits.
Quickly, for defenders:
1. Please research all the things starting from the sample, this is the confirmed CVE-2026-21509 0day intinal attack vector sample, and add detections (currently the detection ratio on VT is pretty low) in the full chain.
2. If you're an Microsoft Office user, please apply Microsoft's official patch or workarounds ASAP msrc.microsoft.com/update-gu…, as now the attacking exploit is well known so attacks are expected to increase rapidly.
For me, there's more work to do, including an EXPMON update for determinate detection against this 0day exploits. In the meantime, if you see the "activex compatibility shellexplorer registry key accessed" Indicator on EXPMON, please be cautions because that's likely the 0day sample/variants.
#CVE-2026-21509 #expmon #0day #zeroday #exploit #threatintel
vessial retweeted
NVIDIA released a CUDA accelerated RTL logic simulator:
github.com/NVlabs/GEM
Open source stm8 idapython disassembly plugin at github.com/vessial/stm8 , first version, you can update opcode parser if you needed
vessial retweeted
Developed by IDApython, IDA Pro built-in plugin only support 162 opcodes by default. I developed idapython plugin it can be fully supported 397 opcodes, almost 800 instructions statements, just diff with IDA built-in plugin disassembled results, it will open source soon.
Now IDA Pro can fully support TI TMS320C28x Series DSP instruction Sets,including C28x assembly language instructions/fpu/fpu64/Viterbi,Complex Math and CRC unit(VCU)/VCRC/VCU-II/Fast Integer Division Unit(FINTDIV)/Trigonometric Math Unit (TMU)/Control Law Accelerator(CLA).
vessial retweeted
All the details about this vuln and much more will be revealed tomorrow by us (me, @bzvr_, @kucher1n) during our talk “Operation Triangulation: What You Get When Attack iPhones of Researchers” at #37c3 (14:45 CET). There will also be a live stream. fahrplan.events.ccc.de/congr…
Qemu fully Tesla model 3 EMMC dump emulation up, this is after 2021 model based on AMD Ryzen APU @greentheonly
vessial retweeted
RCE in MediaTek basebands: in today's blogpost, we explore more CSN1 parsing bugs, this time in MTK's basebands running on MIPS16e2, and analyze how to exploit heap overflows in this baseband OS! labs.taszk.io/articles/post/…
vessial retweeted
Our #Pwn2Own 2021 exploit and a new attack surface about the reparse point memory corruption EoP bug were disclosed in our #BlackHat Europe talk "From Logic to Memory: Winning the Solitaire in Reparse Points". Thanks to @tiraniddo's great previous work :) blackhat.com/eu-21/briefings…
vessial retweeted
Replying to @msftsecresponse
@msftsecresponse said that CVE-2021-1675 is just LPE, so maybe this 100% stable RCE as SYSTEM on Domain Controller is another one? 🤔


































