Nvidia Unveils AI Safety Platform to Prevent Agent Hacks

Nvidia has introduced an open software platform and reference system designed for agentic artificial intelligence safety. The Nvidia Open Agent Safety Platform is aimed at bolstering 'AI security from agent testing to deployment,' according to a Monday (Sept. 28) news release.

The new platform includes Nvidia OpenShell secure runtime software that establishes boundaries for agents running on central processing units (CPUs). As agents take on more work across more systems, enterprises need an enforceable boundary outside of the model and agent harness, the release said.

Speaking to reporters before Monday's announcement, Justin Boitano, Nvidia's vice president of enterprise AI, said the technology would have prevented a hack by OpenAI agents on the AI company Hugging Face in July had it been available at the time. 'From what we know, this new security platform could have stopped the breach,' Boitano said.