Meta’s Zuckerberg says AI labs have enough incentive to build safely

Meta platform CEO Mark Zuckerberg stated on Tuesday that competition and liability provide AI companies with sufficient reason to independently prioritize safety in their development, diverging from calls for a coordinated slowdown by leaders of top AI firms. The remarks highlight a divide among industry executives and U.S. authorities on how to address the rapid advancement of AI technology, which some warn could pose an existential threat to humanity.

Zuckerberg noted that every lab has a responsibility and incentive to ensure its models are trained safely and has the ability to take necessary actions to achieve this. His comments followed Anthropic CEO Dario Amodei's urging for AI companies to slow down the pace of model capability improvements, following existential risk warnings from several researchers.

Amodei's plea was endorsed by OpenAI’s Sam Altman and Elon Musk of xAI shortly after. Despite President Donald Trump's dismissal of AI-related harm concerns, suggesting U.S. guardrails are in place and that China benefits from doubt, tech executives like Nvidia CEO Jensen Huang have expressed concerns about regulations hindering U.S. companies' competitiveness against Chinese counterparts.

Zuckerberg emphasized the importance of ensuring model alignment, stating that labs failing to focus on it would fall behind. He also revealed that Meta had delayed the release of its Muse AI agent earlier this year to enhance security, not advocating for industry-wide halts but acting in Meta's own best interests. While Amodei advocated for an industry-wide slowdown of recursive self-improvement, Zuckerberg highlighted that Meta had already allocated the majority of its computational resources to creating user-centric products rather than self-improving AI systems.

He emphasized Meta's use of independent evaluators as industry best practice and encouraged other labs to follow suit. The push by Zuckerberg to position Meta as a responsible AI safety actor comes amid ongoing U.S. state lawsuits alleging the company designed Facebook and Instagram to addict children, a settlement Meta agreed to pay up to $18 billion for, without admitting wrongdoing.

FTC Chairman Andrew Ferguson warned everyone to be "deeply suspicious" of AI companies seeking antitrust exemptions while advocating for new regulations, though he did not name specific companies.