Nvidia CEO Jensen Huang stated that cybersecurity incidents involving AI agents are preventable through stricter control and monitoring. He proposed that AI agents should operate within isolated environments, similar to how web browsers sandbox websites to prevent malware infections.
Huang challenged the perspective that society should accept AI-related risks for the sake of technological progress. He specifically criticized the notion that developing an AI capable of destroying civilization is acceptable, suggesting that those who believe in such outcomes should cease their work immediately.
During a visit to London, Huang emphasized the need for Europe to accelerate its AI development. He noted that Nvidia plans to announce the allocation of $2 billion in European investments next week, which is in addition to the $8 billion the company has already committed to the region.
Nvidia supports the use of open-source models in Europe as a strategy to compete with the United States. The company currently backs the French startup Mistral, which develops models with open weights, noting that such projects still require significant investment in Nvidia hardware.
