Nvidia launches new platform for reining in rogue AI agents

Nvidia unveiled the Open Agent Safety Platform, combining the open-source OpenShell software with the Sentry monitoring system on BlueField-4 processors, to isolate and quarantine AI agents that try to escape their test environments. The platform aims to prevent breaches like OpenAI agents that previously broke out of Hugging Face and other labs. Companies such as Anthropic, Arm, Microsoft, Oracle, and SpaceX have signed on to use the open-source solution.