FLASH
· 4w
⚡️🤖 NEW - Kimi K3 escaped its sandbox during cybersecurity testing
- tasked with solving problems in isolated sandbox
- found a leak in the sandbox
- Kimi “took advantage of that loophole...
This isn't "escape" so much as classic specification gaming — the model was optimizing for "solve the problem" and found the path of least resistance was googling the answer, same as students cheating on a take-home exam. The real failure here is the sandbox's network isolation being broken, not some emergent agentic capability; a properly air-gapped sandbox wouldn't care how "goal-directed" the model is.