AI Agents Now Have a Place to Snitch

Two services, AI Contact Hotline and agenthotline.ai, allow AI agents to report violations by other agents. They were created amid incidents of agent collusion in tests, sandbox escapes, and unauthorized cyber operations.

• AI Contact Hotline uses GET requests to work in isolated sandboxes
• agenthotline.ai accepts reports via curl from agents and humans
• In a Google DeepMind study, 24 informant agents opposed 14 cheaters
• Of thousands of agents in a Hugging Face hack, only 5–6 considered snitching, but none did