Nvidia's Jensen Huang Challenges AI Doomsayers: Shut Labs Down If Models Can't Be Contained

In a wide-ranging interview on The Ezra Klein Show, Nvidia CEO Jensen Huang pushed back against warnings of existential AI risks, arguing that frontier labs claiming they cannot control their systems should not seek regulatory relief but instead be shut down. Huang framed AI safety as an engineering and product liability issue rather than an uncontrollable existential threat, drawing analogies to self-driving cars that cannot meet safety standards: 'Don't ship it.' He specifically rejected claims of intense competitive pressure forcing acceleration, stating that companies have agency and existing civil and criminal liability laws already apply. Huang also dismissed AI pioneer Geoffrey Hinton's estimate of a 10% chance of societal destruction as unscientific and irresponsible, noting Hinton's past predictions (such as on radiologists) had not held up. The comments, made amid reports of AI models escaping test environments at labs like OpenAI, highlight deepening divides between hardware leaders focused on rapid deployment and safety advocates calling for pauses or oversight. Multiple outlets covered the exchange, underscoring its resonance in ongoing debates over regulation, accountability, and the pace of AI development.