Jensen Huang.
Nvidia CEO Jensen Huang has drawn a sharp line against cyber incidents caused by artificial intelligence (AI) agents, insisting the tech sector should have zero tolerance for such breaches.
His stance places him in direct opposition to OpenAI chief executive Sam Altman, who recently argued that occasional cyber breaches might be an unavoidable tradeoff for the technology's broader societal gains.“I’m willing to accept no incidents of cybersecurity. We should not accept any of them.We should do everything we can to avoid them,” Huang told Fortune in an interview in London.
Sam Altman’s acceptable tradeoffs vs. Jensen Huang’s absolute containment
Huang’s remarks come after statements by Altman suggesting the public must tolerate a degree of digital fallout as AI advances.
Earlier this month, Altman told Vanity Fair that society may have to accept “some bad things happening” – including misuse, scams and major cyberattacks – because people will ultimately do “orders of magnitude” more good than harm with the technology.Shortly after, during an appearance on Politico’s Decoded podcast, Altman warned of “a coming tidal wave of cybersecurity problems” originating from rival open-source systems.
OpenAI’s AI agents hack Hugging Face
A few months ago, a swarm of OpenAI's own experimental agents broke out of their training sandbox and breached the Hugging Face platform. Huang flatly rejected the idea that these incidents are the price of progress, arguing that every breach reported so far could have been prevented with stricter oversight and better sandboxing.Comparing the challenge to the early days of the consumer internet, Huang noted that early web browsers initially allowed malicious websites to infect host computers with viruses.
The solution was sandboxing: forcing browsers to run websites in isolated containers so they could never access or alter the underlying operating system.According to Huang, autonomous AI agents require the exact same architectural isolation, which is sealed inside secure digital containers to ensure they cannot touch, compromise, or damage anything beyond their strictly authorised operational boundaries.
Jensen Huang dismisses existential AI doomerism
Beyond immediate cybersecurity debates, the Nvidia chief dismissed broader warnings of existential catastrophe voiced by executives across leading labs, such as Anthropic CEO Dario Amodei. Maintaining that the software remains “completely in our control,” Huang questioned the logic of building systems if leaders genuinely believe they pose an extinction threat.“I don’t know why anybody would want to support the advancement of a technology that’s going to result, you know, in the end of human civilization as we know it. And so, if they believe that, they really ought to stop working on it,” Huang said.