- Published on
Nvidia CEO Jensen Huang: Extreme measures could include cutting off chip supply if customers develop unsafe AI software
- Authors
- Name
- DeepHorus
Nvidia CEO Jensen Huang said in an interview with Bloomberg reporter Emily Chang in London that if a customer is determined to be developing unsafe or irresponsible artificial intelligence software, Nvidia will stop supplying chips to it. He said that cutting off supply is still a "hypothetical question" at present. Before taking extreme countermeasures such as cutting off supply, Nvidia would inevitably first engage in in-depth communication and dialogue with the relevant customer.
This statement came against the backdrop of an ongoing escalation in the global debate over AI safety. Executives at companies such as Anthropic have suggested that the entire industry should slow the pace of AI development. Huang clearly rejected this. He believes that the various current safety concerns are actually "growing pains" in the process of the industry's rapid expansion.
Huang added that Nvidia does not run other people's companies, and its technology is general-purpose technology. Therefore, regarding how other companies use Nvidia's technology, the judgments and degree of control the company can exercise are actually very limited. This also means that cutting off supply is not a routine measure that Nvidia would easily resort to; rather, it needs to be based on communication and judgment.
In Huang's view, public anxiety about AI safety risks somewhat exceeds the actual situation. He firmly believes that AI software can be effectively governed and safely deployed. He also said that many young companies, as they expand in scale, will gradually mature in their accountability and sense of responsibility.
On the specific path for addressing AI safety hazards, Huang proposed a technical governance approach of "using AI to monitor AI." He argued that the industry should not slow down by giving up eating for fear of choking, but should instead accelerate the integration of prevention, control, and monitoring technologies into AI systems. He said bluntly, "AI itself is the best monitoring tool."
To honor this commitment, Nvidia announced that it will donate $250 million in advanced computing power to UK AI safety research over the next five years, equivalent to approximately RMB 1.676 billion. This computing power will support multiple top UK academic institutions, including the University of Oxford. The related research will deeply explore AI safety protection and agent control methods.
From Huang's statements, Nvidia is simultaneously keeping two paths on the AI safety issue. One is to use chip supply as a countermeasure option in extreme cases; the other is to support safety research with computing power donations. He rejected calls for the entire industry to slow down and advocated accelerating the integration of monitoring and prevention and control technologies. How the relevant stance will be implemented going forward remains to be seen.
