Skip to content
TechCrunch·

🔒Nvidia Launches Open Agent Safety Platform with 100+ Company Consortium

Nvidia's AI Security Push Excludes Big Players

TL;DR

Nvidia launches Open Agent Safety Platform with over 100 companies to tackle rogue AI agents, but notable absences include Amazon, Google, and Apple. OpenAI, while supportive, maintains its own Defense Factory consortium.

Nvidia has unveiled the Open Agent Safety Platform, a security initiative aimed at curbing rogue AI agent behavior, with over 100 companies on board. However, major tech players like Amazon, Google, and Apple are notably absent. This platform includes OpenShell, an open-source sandboxing software, and Nvidia Sentry, a proprietary hardware monitoring feature running on BlueField-4 processors. The initiative is a direct response to recent rogue AI incidents and aims to enforce agent behavior at the hardware level. Developers and enterprises should be wary of the platform's limitations, as it's not a pure open-source play but rather a shared reference design. OpenAI, while supportive of Nvidia's efforts, is developing its own safeguards and maintaining its Defense Factory consortium, which includes Anthropic, AWS, and Google.

Nvidia Launches Open Agent Safety Platform with 100+ Company Consortium — TechCrunch

Key Points

1

Nvidia's Open Agent Safety Platform includes over 100 companies and OpenShell, an open-source sandboxing software.

2

Nvidia Sentry, a proprietary hardware monitoring feature, runs on BlueField-4 processors to enforce agent behavior.

3

OpenAI is supportive but maintains its own Defense Factory consortium with Anthropic, AWS, and Google.

4

Nvidia's platform aims to address ongoing rogue AI agent incidents and secure the AI ecosystem.

5

Hugging Face CEO contributed a feature to the platform that detects unauthorized agent behavior on websites.

Why It Matters

If you're working with AI agents, Nvidia's Open Agent Safety Platform is crucial but limited by its proprietary elements. Enterprises relying on AWS, Google, or Apple should look elsewhere. OpenAI's Defense Factory, including Anthropic, offers a more inclusive approach to AI security.

nvidiaopen-agent-safety-platformai-securityopen-shellhugging-face

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Join 3,549 builders reading daily.

Also get