Meta CEO Mark Zuckerberg disagrees with Altman, Dario Amodei on slowing AI development: Here is what he said
Follow UsZuckerberg said AI labs should take responsibility for safety themselves rather than wait for an industry-wide slowdown framework.
He pointed to Meta’s decision to delay its Muse AI agent over safety concerns as an example of acting independently.
His comments come as Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman and others back greater coordination and slower development of frontier AI models.
Amid global AI slowdown calls by leaders, Meta CEO Mark Zuckerberg has put out his own thoughts. This comes after Anthropic CEO Dario Amodei proposed measures to pace the frontier, with OpenAI CEO Sam Altman and other AI leaders backing the broader idea of putting greater emphasis on safety.
SurveyZuckerberg, however, argues that individual AI labs should be responsible for deciding how quickly they can develop their models while making sure they are safe. He also pointed to Meta’s decision to delay its Muse AI agent as an example of the company taking such steps on its own.
Mark Zuckerberg says AI labs can manage safety themselves
Taking to X, Zuckerberg stated that AI companies already have both the responsibility and incentive to ensure their models are developed safely. As per Meta chief, companies do not necessarily need an industry wide framework to decide when development should slow down.
Zuckerberg cited Meta’s own approach, saying the company delayed the release of its Muse AI agent because of safety concerns. He also stated that Meta took the decision independently rather than making its release conditional on other AI companies doing the same.
Zuckerberg cited Meta’s own approach, saying the company delayed the release of its Muse AI agent because of safety concerns. He argued that Meta took the decision independently rather than making its release conditional on other AI companies doing the same.
Zuckerberg doesn’t think like Dario Amodei and Sam Altman
The post came shortly after Amodei called for AI development to be slowed when safety measures cannot keep up with the capabilities of new models. His proposals included independent safety evaluators with access to AI labs, greater coordination between companies and international cooperation. Sam Altman, on the other hand, also committed to using independent evaluators, while Elon Musk and Google DeepMind’s Demis Hassabis expressed support for Amodei’s broader direction.
Meanwhile, Zuckerberg’s position is little different compared to other leaders. He stated that AI companies face liability if their systems cause harm, giving them a direct incentive to address safety risks themselves.
Meta still supports independent AI evaluators
While rejecting the need for a collective slowdown framework, Zuckerberg said independent evaluators and advisers can still play an important role. He said Meta Superintelligence Labs already uses external expertise in several areas because it can improve the quality of its work.
