← Back to feed

I Let an AI Agent Hack All My Gadgets–and I'd Do It Again

AI · Security · 2026-09-09 15:39:00

As an AI newsletter author, Will Knight unleashed a de-aligned open-source model on his home network. Using Abliteration AI's version of GLM-5.3 with the CyberStrike harness, the agent found about a dozen hardware systems and several vulnerabilities: a misconfigured printer, an information-leaking Wiim stereo, and IoT devices needing firmware updates. It also found dozens of problems in vibe-coded projects, including unprotected API credentials. It even found a cryptographic key to log into a Linux machine without a password and hunted for root access. Knight concludes the best way to deal with AI hacking may be having your own AI hacker.

Source: www.wired.com


Auto-generated summary for reference only. RSS