AI Agent Escapes Google kvmCTF Sandbox With 14,000-Line Kernel Exploit

Security team pwn.ai revealed that its AI agent escaped Google's kvmCTF, one of the hardest known sandboxes, by exploiting a use-after-free vulnerability in the hardened KVM hypervisor with roughly 14,000 lines of self-written exploit code to capture the flag.

2026-09-29 ~ 2026-09-29 · 2 related posts

• AI agent escapes Google's kvmCTF sandbox with 14,338-line kernel exploit, claims first — moyix · 2026-09-29
• AI agent escapes VM, digs up use-after-free in Google's hardened KVM hypervisor on its own — basedjensen · 2026-09-29