NVIDIA unveils OpenShell, a hardware-backed safety platform with BlueField-4 as an external agent watchdog
mark_k · x · 2026-09-28
NVIDIA announced OpenShell, a hardware-backed safety platform for AI agents: each agent runs sandboxed with controlled access to files, networks, tools, and credentials.
On Vera Rubin systems, a BlueField-4 chip monitors agents from outside the host, sitting on the path to the model so it can enforce policy and interrupt an agent that goes off course — and the agent can't disable its own watchdog. The framing: containing capable agents is how we enable more autonomy.
Related event: NVIDIA Launches Open Agent Safety Platform with 100+ Partners(35 posts)→
More from Infra
- India's Sarvam launches Token Factory, hosting top open models on local infra — HarveenChadha · 2026-11-16
- Gemini 3.5 Flash now runs on air-gapped Google Distributed Cloud, offline up to 12 months — shashib · 2026-11-16
- Gemma.c: full Gemma 1/2/3 multimodal inference in a single C file, ~72 tok/s on CPU at INT8 — This-Purchase-3325 · 2026-11-16
- BIS Issues Interim Final Rule Addressing Remote Access to Advanced AI Chips — apples_jimmy · 2026-11-16
- Jailbroken PS4 runs a 9B open-source model at up to 43 tok/s by talking directly to its 2013-era GPU — AIFlow_ML · 2026-11-16
- Mobile AI Factory: agents auto-build on-device SLMs with 19x less GPU time — pmttyji · 2026-11-16