CTF Player @ L3H_Sec / OIer / System Security / MSRC MVR 2026 #27 / Exploitation / CS PhD Student / @DiffractAI
Joined June 2020
- Tweets34
- Following138
- Followers274
- Likes113
Qinrun Dai retweeted
Glad to see 86 Windows CVEs reported by @DiffractAI fixed this October. We’ll gradually share more of our work across Windows and other platforms.
Microsoft’s October 2026 security updates fixed 86 Windows CVEs reported by Diffract, spanning the Windows kernel, file systems, storage stack, device drivers, and other core components.
We’ll continue digging into underexplored Windows attack surfaces while expanding Diffract’s research to other operating systems. More research and technical write-ups are coming.
msrc.microsoft.com/update-gu…
Qinrun Dai retweeted
We’ve published our analysis of CVE-2026-97798, a Windows Faultrep TTD option-injection vulnerability discovered and reported by Diffract. It allowed a standard user to execute native code as SYSTEM through a trusted broker.
Technical details and demo:
diffract-ai.com/research/cve…
It's crazy, after seeing Anthropic release new models for cyber, I wanted to check them out, but the CVP portal suddenly showed my verified status as not verified again. What's even weirder is that after going through Persona, my account just got suspended for no reason... I have no idea what happened.
Qinrun Dai retweeted
This month, Microsoft’s @msftsecresponse fixed 80 CVEs discovered and reported by me and the outstanding collaborators on our team, including wgg, npc0vo, and 2st. MSRC’s response and remediation were remarkably swift.
#bugbounty #cybersecurity #MSRC #security
Security updates for September are now available: msft.it/6018SZEg0.
Alongside this month's release, we're expanding machine-readable Vulnerability Exploitability eXchange (VEX) coverage to all Microsoft-assigned CVEs, providing customers with more consistent, machine-readable security information to help understand exposure and prioritize risk.
Learn more about this latest milestone in our transparency efforts: microsoft.com/en-us/msrc/blo…
Diffract (developed by @HototogisuTian @2st___) is an AI vulnerability research agent that goes beyond LLMs—we focus more on the nature of bugs than on AI engineering.
Q2 results:
• 37 Microsoft acknowledged CVE
• #1 on the MSRC Windows leaderboard
• #4 overall on MSRC
• Linux kernel bugs too.
We're continuing to work hard to push this further :)
msrc.microsoft.com/leaderboa…
#MSRC #Windows #Vulnerability
Qinrun Dai retweeted
Congratulations to all the researchers recognized in the MSRC 2026 Q2 Security Researcher Leaderboard!
This quarter marks the final quarterly points-based leaderboard as we continue the evolution of our researcher recognition program. We are grateful to every researcher who partnered with us and helped strengthen the security of Microsoft products and services through their hard work and dedication.
We’d also like to recognize the top 10 researchers in this quarter’s leaderboard for their outstanding contributions. Thank you for helping protect customers around the world.
🥇Asaf Cohen (XBREACH.AI)
🥈C46F3708A45EF0041BD0A49DDAEA0E25
🥉Ron
4. Thanatos Tian (HKPolyU) & wgg & 2st with Diffract
5. haowei yan
6. fce141fd6b42fcec05c1285b15d8f999
7. Ofek Levin
8. pwn2addr
9. Kim Seung Chan (@mylostchristmas)
10. Jianyang Song
Learn more and view the full leaderboard: microsoft.com/en-us/msrc/blo…
Finally bought my second Claude Max 20x plan. I’m a top AI security research payer—wonder if I really have to submit another form just to get the cyber use restrictions released on a second account?
Another RCE is born. Finding a useful infoleak is 10x harder than the OOB write. Hoping to catch this year's BlackHat :)
I’m honored to be here at #BHUSA 2024 to deliver my talks about 1 click RCE on smart phones and machine learning into kernel on-the-fly quarantine!
What's the feeling of rewriting code based on 4-years-ago toolchain version and 10-years-ago historical projects?
Qinrun Dai retweeted
Boulder faculty and students are pround of being part of the 42-b3yond-6ug team. Now, take a shot at the meaning of the team name.
Congrats to the 7 companies that will receive $1 million each to develop AI-enabled cyber reasoning systems that automatically find and fix software vulnerabilities as part of the #AIxCC Small Business Track! Full announcement: darpa.mil/news-events/2024-0….
Qinrun Dai retweeted
Do you use a virtual machine to browse dangerous links safely? If you use the Chrome browser inside that virtual machine, is it secure enough?
As you might have guessed, the answer is not so much.
We chained six unique CVEs from 2023 listed below.
• Chrome Renderer RCE : CVE-2023-3079
• Chrome Sandbox Escape : CVE-2023-21674
• LPE in guest OS : CVE-2023-29360
• VMware Info Leak : CVE-2023-34044
• VMware Escape : CVE-2023-20869
• LPE in host OS : CVE-2023-36802
Recently it occurred to me the fact that when I lose my status as a student, I will no longer be a "maybe-good" CTF player or a security researcher. I will still need to fill my knowledge with 7x10 hours of study and enthusiasm like I did during my sophomore year.
At first I was used to read LLVM source code a lot to find what methods, class...that I should use . Now with openai and copilot, I can focus on the algorithm design, instead of "wasting" time to read the source. AI is indeed a good helper to quickly learn prior knowledge.
Qinrun Dai retweeted
Uploaded all my Offensive Security & Reverse Engineering (OSRE) course labs (docx) to my repo found below. Most of them have very detailed instructions and should be great to get you started in Software Exploitation. 1/n
#Offsec #SoftwareExploitation #RE
exploitation.ashemery.com/
Qinrun Dai retweeted
How to Reverse and Exploit iOS for BEGINNERS😈
👇My 3 part series👇
Part 1: How to Reverse & Patch iOS Apps
bit.ly/3inJgop
Part 2: Exploiting iOS binaries: ARM64 ROP Chains
bit.ly/3BgvLxt
Part 3: Heap Overflows on ARM64: Spraying, UAF
bit.ly/3gGI7rx
Finally, my TOEFL score is out !!!
No longer tormented by exam cancellations and worrying about hold status !!!