According to Zuckerberg, it is in everyone's interest to ensure that security works in the AI models being developed, but that development does not need to be slowed down by cross-industry regulations.
"We waited months to launch Muse (Meta's AI service) to focus on security. We didn't demand that everyone else do it before we did," Zuckerberg writes on X.
He believes that companies already have incentives to prioritize security in development, but that they are market-based.
“Companies risk significant liability if their models cause harm,” Zuckerberg writes.
Jensen Huang believes that it is up to each company to ensure that its models are safe.
"We should develop products and test them properly. And if they're not ready to release, we should wait and continue testing and further developing them until they're ready," Huang told CNBC.
The statements come after OpenAI's and Anthropic's CEOs, Sam Altman and Dario Amodei respectively, as well as entrepreneur and billionaire Elon Musk, have all warned that AI is now developing at a pace that poses potential risks to humanity.
