Jensen Huang said it is absolutely necessary for Nvidia to cooperate with Chinese AI companies to jointly monitor risks from frontier AI, describing AI as potentially dangerous and calling for dialogue to promote safe AI development. He also said he has been in contact with Chinese researchers.
In a Bloomberg interview in London, Jensen Huang said Nvidia would stop supplying chips to customers developing unsafe AI software, but would communicate before cutting off supply. He stressed Nvidia's technology is general-purpose and its involvement has boundaries; he called AI safety concerns overstated and AI bubble talk exaggerated, said supply-demand imbalance would last for years, and proposed using AI to monitor AI.
In an interview, NVIDIA CEO Jensen Huang said frontier labs must be shut down if they cannot control AI safety. Responding to jailbreak attacks on frontier models from OpenAI, Anthropic, Meta, and Google, he said rogue agents could lead to civil and criminal liability for labs, and that current worries are a distraction from focusing on experimental controllability.
In a nearly two-hour New York Times podcast interview on Sept. 23, 2026, Nvidia CEO Jensen Huang argued that AI companies should not receive antitrust or product-liability exemptions if they seek regulation, reiterated that AI labs must test products and ensure safety before release, said he supports regulation for specific applications like autonomous vehicles, and discussed the OpenAI agent intrusion at Hugging Face, which Nvidia agreed to acquire for $13 billion.