Nvidia Unveils Software to Prevent AI Security Incidents
Nvidia has released software designed to prevent AI security incidents, a response to recent breaches by companies like OpenAI and Anthropic.
The 'trust layer' platform aims to insulate AI agents from exposure to other AI systems and the open internet during their existence.
This move comes after warnings from industry leaders about the risk of AI slipping out of human control, including a potential pathway called recursive self-improvement.