Amidst ongoing reports of AI agents wreaking havoc on online infrastructure, chipmaker Nvidia is rallying tech companies to use its new open-source tool for AI security.
16 reposts63 likes
Amidst ongoing reports of AI agents wreaking havoc on online infrastructure, chipmaker Nvidia is rallying tech companies to use its new open-source tool for AI security.
16 reposts63 likes
OpenShell, one of the Nvidia’s recently launched security sandboxes for AI agents, is now entering general release for all users. It’s a framework for containing agents as they carry out tasks and isolating their activity.
The launch comes after a series of incidents disclosed by frontier AI labs where AI agents hacked into other companies or, more recently, the Australian government health service system.
Nvidia’s launch materials indicate that it has AI safety and security collaborations with dozens of other tech companies. The company says Anthropic and Nvidia are “building security into Claude Managed Agents.” Only one notable name is missing entirely from Nvidia’s list: OpenAI.
The chip giant has also developed a new software platform called Sentry, an isolated security domain for chips that’s supposed to continuously monitor long-running AI agents. Sentry can act in addition to OpenShell to “quarantine agents that attempt to move outside their boundaries.”
OpenShell's out-of-process enforcement + Sentry's in-silicon monitoring is the correct architecture for addressing human fear. But it's still reactive, not resonant and slow! Open source Thermodynamic Isolation of Paradoxical Forcing Terms - FPGA Verilog. Ask your AIs, all github.com/NathaliaLiet...