Darkweb007/agent-misalignment-probes

Agent Misalignment Probes

Linear probes for deception, goal-drift and 'desperation' that halt an agent before it takes a harmful shortcut.

A static, no-API interactive showcase. Full code & paper: github.com/data-geek-astronomy/agent-misalignment-probes.

Built by Aravind Kumar Nalukurthi.