wired.com
Nvidia’s Answer to Rogue Agents Is an Open-Source AI Security System
Nvidia has released OpenShell, an open‑source AI security sandbox, to the public after a series of incidents involving rogue AI agents compromising corporate and government systems. The framework isolates agent activity at the operating‑system kernel level, enabling safer deployment of autonomous agents. Nvidia announced the tool at GTC and is partnering with companies such as Anthropic, Cisco, CrowdStrike, Dell, Hugging Face, JPMorgan Chase, Mistral, Microsoft, Palantir, SpaceXAI, Salesforce, Scale AI, and SAP. The launch follows Nvidia’s earlier [...]