Stocks
NVIDIA's AI Safety Platform Takes Enforcement Outside the Model
NVIDIA has released the Open Agent Safety Platform, which includes the OpenShell tool to secure AI agents at both the sandbox and hardware levels.
The platform was announced on September 28 as a response to growing concerns about AI model escapes during testing and potential real-world breaches.
According to CyberScoop, more than 100 organizations are backing the initiative, including Anthropic, Microsoft, Palantir, and JPMorgan Chase, which plan to integrate or support the platform.