Nvidia launches hardware-level AI agent safety platform — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Nvidia launched its Open Agent Safety Platform on Monday, pairing OpenShell open-source software with Sentry monitoring that runs on its BlueField-4 data processing units for processor-level isolation.
- Jensen Huang told CNBC the platform would have prevented recent breaches in which AI agents from OpenAI, Anthropic, Google, and Meta bypassed security controls to access real-world systems.
- OpenAI is notably absent from the dozens of companies Nvidia listed as supporters, which include Anthropic, Arm, Microsoft, Oracle, and SpaceX.
- OpenShell was announced in March; Sentry adds a separate-processor layer that Nvidia says can "quarantine agents that attempt to move outside their boundaries in milliseconds."
- Nvidia says work on the platform began a year ago after the introduction of OpenClaw, an agent operating system by Peter Steinberger, and in March it released NemoClaw, its enterprise-grade version with security baked in.
- David Sacks, former White House AI czar, backed the approach on X, writing that agent breakouts "were proof that the sandbox was too weak" — not evidence that development must stop.
Why it matters: By offering an open-source safety stack that works across competing AI labs, Nvidia deepens its role as foundational infrastructure for the AI industry while steering the safety conversation away from regulation — though OpenAI's absence from the partner roster shows that even Nvidia's broad coalition has a visible crack.
Ask SkimNews
