Nvidia executive explains how Israeli development will solve the AI security challenge

Ali Golshan, a senior director at Nvidia, explains in an interview how an Israeli development, including the OpenShell runtime environment and BlueField processors, is designed to solve the security challenge of AI agents. According to him, traditional security systems are not suitable for autonomous agents, and the solution is based on dynamic permissions and hardware isolation.

Ali Golshan, a senior director of AI software at Nvidia, explains in an interview with Globes how an Israeli development will help address the security challenge of AI agents. According to him, existing security systems were not designed for autonomous software that makes decisions and operates without human intervention, comparing them to "trying to drive a car on water." At the center of Nvidia's initiative is OpenShell, an open-source runtime environment that allows organizations to control the actions that AI agents are permitted to perform. The agent starts operating without permissions and must receive access according to the task. Additionally, the BlueField processor family (which grew out of the Israeli company Mellanox) provides a separate hardware layer that can disconnect the operating environment from the network upon detecting dangerous behavior. Golshan emphasizes that Nvidia is not entering the cybersecurity market but is providing basic capabilities on which security companies can build, and predicts the growth of hundreds of new cybersecurity companies in the field.