OpenAI partnered with researchers to demonstrate Codex autonomously exploiting a Samsung TV, escalating from browser code execution to root by discovering and chaining vulnerabilities without human-provided exploits. The AI analyzed KantS2 firmware source, identified a physical-memory mapping primitive in the ntksys driver, and adapted to Samsung's execution constraints. This demonstrates significant autonomous AI capability in real-world hardware vulnerability discovery and exploitation.
Models
Codex Hacked a Samsung TV
OpenAI's Codex autonomously discovered and chained vulnerabilities on Samsung TVs, escalating from browser code execution to root access without human-provided exploits—demonstrating significant AI capability in real-world hardware exploitation.
Thursday, April 16, 2026 12:00 PM UTC2 MIN READSOURCE: Hacker NewsBY sys://pipeline
Tags
models