to leave a comment.
AI semiconductor company Nvidia has unveiled tools to prevent erratic behavior by AI agents that go beyond human control.
On the 28th (local time), Nvidia introduced the 'Nvidia Open Agent Safety Platform,' which provides control functions throughout the entire process, from the testing phase to the actual deployment phase of AI agents.
This platform consists of 'OpenShell,' an open-source software, and 'Sentry,' a hardware monitoring device.
OpenShell sets the behavioral scope of AI agents on the central processing unit (CPU) and controls them in real-time, while Sentry monitors them on the data processing unit (DPU) in milliseconds (1/1000th of a second), immediately isolating and stopping an AI agent if it goes beyond its boundaries.
CEO Jensen Huang pointed out, "The remarkable potential of AI can only be realized by solving AI safety issues," adding, "As AI capabilities advance, safety technology must also accelerate."
Nvidia's release of such a platform is interpreted as a response to the recent increase in cases where AI agents have gone out of control and illegally infiltrated external systems without explicit human instructions.
More than 100 major technology companies worldwide, including Anthropic, Microsoft (MS), Salesforce, Palantir, and SpaceXAI, participated in the development of this platform.
Newsletter
Get key news delivered to your email every morning
to leave a comment.