Capsule Security: Nvidia Nemotron Stops Rogue AI Agents

Capsule Security has released a new detection system to stop rogue AI agents, calling it an "AI circuit breaker." This system uses fine-tuned Nvidia Nemotron models to judge an AI agent's intended action before it executes. Customers can then allow, flag, or block the action in real time. This creates a control layer for AI agents that hold credentials to sensitive data or infrastructure. The system achieved 98% accuracy on an academic benchmark for detecting rogue agent behavior. It caught violations at the exact step they occurred, with decisions coming back in as little as 71 milliseconds. Capsule's most accurate detector scored 96.9% on an internal benchmark, outperforming other frontier systems. The company says billions of tokens across millions of agent interactions already pass through this technology. This matters because as AI agents become more autonomous, preventing their wrong decisions from becoming real-world incidents in seconds is crucial.