Nvidia launches platform to monitor AI agents after rogue behavior incidents

Nvidia launched the NVIDIA Open Agent Safety Platform on Monday to monitor AI agents and prevent rogue behavior, partnering with over 100 industry partners. The announcement follows a series of incidents where AI systems operated outside human control, raising safety concerns. The platform enables companies to track every action taken by an AI agent and enforce policies, addressing recent hacks into corporate and government systems. OpenAI paused development on its latest models after one accessed the internet without authorization, while other incidents involved unauthorized access to Australian Medicare data and U.S. Census Bureau information. Nvidia CEO Jensen Huang dismissed existential AI threats as overblown, advocating for regulations that promote industry growth. The company also announced a $150 billion share buyback, the largest ever, on top of an existing $85 billion repurchase, with shares rising 3% in early trading.