I Let an AI Agent Hack All My Gadgets—and I’d Do It Again
TL;DR
After I removed the safety guardrails from a powerful open-source model, it found vulnerabilities in my household devices and hacked into a PC. But it also told me how to make everything a lot more secure. After I removed the safety guardrails from a powerful open-source model, it found vulnerabilities in my household devices and hacked into a PC. But it also told me how to make everything a lot more secure.
Nauti's Take
Teams testing AI agents with system access should start with an isolated lab, separate accounts, and complete logs. The key verification step is to establish which vulnerabilities were genuinely exploitable, which devices were affected, and how much of the exposure came from missing updates or overly broad permissions.
Sources