On the 28th local time, Nvidia released an AI security platform called the “Open Agent Safety Platform”. The system consists of OpenShell open source software and a Sentry reference system design, which can set safer operating boundaries for AI agents. Nvidia said the platform could have prevented an OpenAI model attack in July this year from hugging the face. OpenShell can provide a security boundary when the agent is running. Once an AI agent attempts to access an unauthorized file, initiate an abnormal network connection, or execute a dangerous command, it will be denied, recorded, or even quarantined. The software can be deployed on Nvidia hardware or extended to third-party platforms such as Anmou and Intel. Sentry runs on Nvidia hardware. It is independent of the agent's software operating environment and continuously monitors the agent's behavior. If an AI agent tries to cross the boundary, it can isolate abnormal agents at a speed of milliseconds. Analysts believe that Nvidia is trying to further embed AI security capabilities into AI infrastructure to reduce the security risks of autonomous agents through permission control and abnormal isolation. This also means that Nvidia's business reach is being further extended from AI chips to the field of software and security infrastructure.

Disclaimer:Webull uses external vendor Google Translation Service for news translations where we endeavour to ensure these are correct, however, we recommend that you please double-check this information accordingly. Webull is not responsible for translation errors or issues.