India, Oct. 8 -- Securing autonomous AI agents is now an infrastructure problem as businesses move from AI chatbots to agents capable of accessing credentials, databases and core workflows. NVIDIA has launched the NVIDIA Open Agent Safety Platform to put enforceable boundaries around these systems from testing to deployment.

On September 28, 2026, NVIDIA officially announced the Open Agent Safety Platform to address growing concerns around AI agents moving outside their defined boundaries and interacting with systems in unauthorised ways. The platform combines an open-source software runtime with an out-of-band hardware watchdog layer.

NVIDIA OpenShell is an open-source secure runtime at the core of the software stack, now broadly avail...