Nvidia engineer on Open Agent Safety Platform: verifiable, monitorable agentic AI
LuizaJarovsky · x · 2026-11-01
AI governance researcher Luiza Jarovsky interviewed Ali Golshan, a senior engineer behind Nvidia's new Open Agent Safety Platform, at GTC Berlin. The platform spans application, runtime, and infrastructure layers to make policies verifiable and agents continuously contained and monitored; OpenShell is released under Apache 2.0. Topics included emerging agentic risks, governing evolving capabilities, and privacy challenges. Context: recent incidents like the OpenAI/Hugging Face event and leaked PHP vulnerabilities. Jarovsky came away optimistic that agentic safety is fundamentally a solvable engineering problem.
Related event: Nvidia Unveils Open Agent Safety Platform for Verifiable AI Agents(2 posts)→
More from Infra
- Local STT shootout: Nemotron 3.5 Transcribe beats Fable on noisy audio — CryptoxPathy · 2026-11-01
- Old PC relabeled as an "AI system" sells for €5,400 after sitting at €750 — sernamenotdefined · 2026-11-01
- Turning AI data centers' waste heat into tea and saunas? A cozy proposal — emax · 2026-11-01
- Running a 22B video model on a 14 GB Vega 7 iGPU: patches, tricks, timings — shirosoro · 2026-11-01
- Intel EMIB packaging yield at 60-65% vs TSMC CoWoS's 98%, report says — rickasaurus · 2026-11-01
- SpaceX guides to $35B quarter as Grok Bot hits 800K paid weekly users — elonmusk · 2026-11-01