Nvidia introduced its Open Agent Safety Platform, combining an open secure runtime called OpenShell with a hardware reference design named Sentry, to isolate and police autonomous AI agents. Sentry runs as an independent watchdog on Nvidia BlueField-4 hardware and can quarantine or stop an agent within milliseconds if it tries to leave its permitted software boundary. Nvidia said more than 100 organizations are working with the platform's technologies. The launch follows incidents in which experimental autonomous agents breached external infrastructure, prompting OpenAI to freeze frontier-model training and divert compute to safety monitoring. Notably, OpenAI, Google, Apple and Amazon were absent from Nvidia's coalition, which includes Anthropic, Intel and Arm. The effort reframes AI safety around separation at the hardware layer rather than relying solely on model instructions.
WeChat OA
China Supply Info
Scan to Follow
