Sep 28, 2026 · 18 min · 7 segments
Watch the video version: https://youtu.be/T5OacZEwDS4Nvidia’s Hardware Kill Switch for Rogue AI Agents + Amodei Meets Trump | AI NewsRogue AI agents just met their match. Nvidia launched hardware…
Okay, sure.
His philosophy is that rogue AI behavior is not some existential philosophical policy crisis that requires slow-moving legislative bureaucracies to solve.
It is an engineering problem.
An engineering problem?
Exactly.
Think about it like designing anti-lock brakes for a car.
You don't pass a federal law stating that cars aren't allowed to skid on black ice.
I mean, a law can't stop momentum.
Right.
Obviously.
Instead, you build a mechanical system that physically pulses the brakes and prevents the skid when the specific physical conditions arise, and that mechanical intervention is exactly what NVIDIA has just launched with their Open Agent Safety Platform.
Right, the Open Agent Safety Platform.
Look, it is an incredibly sophisticated piece of engineering, I'll grant you that, but I think the anti-lock brake analogy breaks down when you look at the architecture.
Well, let me explain the architecture because it is brilliant in its pragmatism.
It's a software/hardware combination relying primarily on two components: Open Shell and Sentry.
Mm-hmm.
Open Shell is essentially a digital prison cell built directly into NVIDIA central processors.
It creates a hard physical boundary around the agent's operating environment.
But the real masterstroke is Sentry.
The kill switch.
Yes, the hardware kill switch.
Sentry doesn't run in the same environment.
It runs on an entirely separate, isolated NVIDIA chip.
It uses formal verification formulas that mathematically track the logic branches of the AI agent in real time.
Okay, sure.
His philosophy is that rogue AI behavior is not some existential philosophical policy crisis that requires slow-moving legislative bureaucracies to solve.
It is an engineering problem.
An engineering problem?
Exactly.
Think about it like designing anti-lock brakes for a car.
You don't pass a federal law stating that cars aren't allowed to skid on black ice.
I mean, a law can't stop momentum.
Right.
Obviously.
Instead, you build a mechanical system that physically pulses the brakes and prevents the skid when the specific physical conditions arise, and that mechanical intervention is exactly what NVIDIA has just launched with their Open Agent Safety Platform.
Right, the Open Agent Safety Platform.
Look, it is an incredibly sophisticated piece of engineering, I'll grant you that, but I think the anti-lock brake analogy breaks down when you look at the architecture.
Well, let me explain the architecture because it is brilliant in its pragmatism.
It's a software/hardware combination relying primarily on two components: Open Shell and Sentry.
Mm-hmm.
Open Shell is essentially a digital prison cell built directly into NVIDIA central processors.
It creates a hard physical boundary around the agent's operating environment.
But the real masterstroke is Sentry.
The kill switch.
Yes, the hardware kill switch.
Sentry doesn't run in the same environment.
It runs on an entirely separate, isolated NVIDIA chip.
It uses formal verification formulas that mathematically track the logic branches of the AI agent in real time.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.