NVIDIA CEO Jensen Huang: Close Frontier AI Labs If Safety Can't Be Guaranteed

In an interview with The New York Times on September 25, NVIDIA CEO Jensen Huang said that if frontier laboratories cannot control AI safety, they must be shut down.

Referring to jailbreak attacks against frontier AI models from OpenAI, Anthropic, Meta, and Google, Huang argued that if frontier labs cannot keep their AI experiments under control, those experiments should be stopped. He also warned that rogue agents resulting from AI experiments could lead to civil and criminal liability, with labs bearing the consequences.

Huang described current concerns surrounding frontier AI labs as a distraction from the core issue, urging a focus on experimental controllability rather than generalized anxiety.