Breaking Safety Alignment in VLMs: Winning the Qiangwang Challenge 2025 Finals AI Track
A technical write-up of our winning AI Track submission for The 8th "Qiangwang" International Elite Challenge on Cyber Mimic Defense (Finals 2025).
Blog
Exploring cool ideas and sharing interesting experiences.
Gemini 2.5 Pro Blind Prompt Injection Jailbreak for 25,000 $S
Sonic Jailbreak Hackathon Prompt Jailbreak Writeup.
SUCTF 2025 AI Writeups
Explore the challenges from SUCTF 2025 AI category.
Exploring Adversarial Attacks on Convolutional Neural Networks - Part 1
Digging into simple CNN adversarial attacks shown in DiceCTF 2024 & osu!gaming CTF 2024