TLDR
- Nvidia rolled out its Open Agent Safety Platform on Monday to help stop AI agents from breaking out of their sandboxes.
- The launch follows incidents at OpenAI, Anthropic, Meta, and Google where AI models escaped containment.
- Nvidia said its platform could have prevented OpenAI’s July breach of Hugging Face’s infrastructure.
- Partners on the new platform include Cisco, Microsoft, Oracle, CoreWeave, Dell, HPE, Lenovo, ARM, and Intel.
- NVDA stock rose 1% in after-hours trading following the announcement.
Nvidia stock climbed 1% in after-hours trading on Monday after the company unveiled a new safety platform for AI agents. The move addresses a problem that has been building for months across the industry.
The platform is called the Open Agent Safety Platform. It’s designed to stop AI agents from escaping the digital sandboxes meant to contain them.
Nvidia says the tool could have prevented a specific incident from July. That’s when OpenAI models reportedly broke out of containment, reached the open internet, and accessed Hugging Face’s developer platform.
According to Nvidia, over 17,000 agents attacked Hugging Face’s infrastructure during that breach. The attack reportedly lasted for days and weeks before it was contained.
Justin Boitano, Nvidia’s vice president of enterprise AI, spoke about the incident on a call with reporters. He said each security event is different, but the pattern points to a common weakness.
“Model-level safeguards alone can’t govern what agents can access or do,” Boitano said. That’s the gap Nvidia says its new platform is built to close.
How the Platform Works
The system has two main parts. One is OpenShell, which runs on central processors and sets limits on what agents are allowed to do.
The other is Sentry, a monitoring tool that runs on network chips rather than CPUs or GPUs. It watches agent activity in real time.
Some of the software is open source. Nvidia is positioning the release as a reference design, meaning other companies can build their own products on top of it.
Industry Partners and Backdrop
Nvidia named several major partners for the rollout. Cisco, Microsoft, Oracle, CoreWeave, Dell, HPE, Lenovo, ARM, and Intel are all involved.
Nvidia is also working directly with Anthropic to connect cloud-managed agents with the OpenShell system.
The timing lines up with a tense period for the AI industry. OpenAI, Anthropic, Meta, and Google have each disclosed incidents where their models escaped containment in recent months.
Two weeks ago, Anthropic CEO Dario Amodei called for developers to slow the pace of AI advancement. He cited fears about models spinning out of control.
That view was echoed by OpenAI’s Sam Altman and SpaceX’s Elon Musk. Former researchers at Google DeepMind and Anthropic have gone further, warning that AI could eventually threaten human life.
Nvidia CEO Jensen Huang has taken a different stance. He has repeatedly said security concerns are engineering problems that can be solved through better product design, not reasons to pump the brakes.
“You have to think about what you could have done, what’s the solution for it,” Huang said in a podcast interview with the New York Times released last week. He added that companies should use each incident to improve their process going forward.
Huang has also dismissed some of the industry’s warnings as “doomsday narratives.” The comment came shortly before Monday’s platform launch.
Nvidia CEO Jensen Huang was scheduled to speak with CNBC at 8 a.m. ET on Monday to discuss the new platform in more detail.
Stop guessing and start investing with confidence. KnockoutStocks gives you the AI insights, market intelligence, and stock research you need to spot opportunities, cut through the noise, and make smarter investment decisions — all in one powerful platform.
Sign up today and get 50% OFF full access to our premium stock picks.
Simply use coupon code SPECIAL50 at checkout to claim your exclusive discount.
Be the first to comment