Meta Platforms Inc (NASDAQ: META) Chief Executive Officer Mark Zuckerberg and the company's AI head, Alexander Wang, have entered the debate over the pace of artificial intelligence development. The two executives argue that AI labs should continue pushing forward with model capabilities while simultaneously building stronger protective measures around increasingly powerful systems. Their comments appeared on X, as the industry wrestles with whether capability expansion should be temporarily paused until more progress is made on AI safety — the discipline of ensuring AI systems reliably adhere to human intentions and values.

Meta shares closed higher for a third consecutive session on Tuesday, accumulating a 4.2% gain over that stretch, with a modest uptick in after-hours trading. On Stocktwits, retail investor sentiment for the stock was last seen at "extremely bearish."

Zuckerberg argues labs already have strong incentives to build safer AI

Zuckerberg stated that AI companies already face significant commercial and legal motivations to develop models that are safe and aligned with user interests. "People will not want to use agents that are not aligned with them and don't do what they ask," Zuckerberg wrote. "Any lab that does not focus on alignment will fall behind." He characterized alignment as emerging into a crucial competitive capability, rather than merely a constraint on AI development. "My view is that trust and alignment are quickly becoming the most important capabilities distinguishing agents and models," he said. Zuckerberg also noted that Meta previously decided to delay the release of its Muse model by several months in order to address safety and security concerns.

Meta calls for external oversight

Both Zuckerberg and Alexander Wang urged greater involvement of independent evaluators and external oversight as AI models grow more powerful. "Bringing in independent evaluators and advisors is industry best practice," Zuckerberg said, adding that a larger and more diverse ecosystem of assessors would improve the overall state of AI safety work. Wang echoed that sentiment, saying AI labs should maintain "robust governance frameworks" throughout the entire training and deployment lifecycle, including external evaluators and independent monitoring of safety standards for model releases. He added that every lab must operate within the institutional protections of democratic nations, meaning labs would face significant legal liability if their models cause harm.

Meta defends its compute allocation strategy

A key part of both executives' arguments centered on Meta's decision to allocate the vast majority of its computing power toward serving users, rather than accelerating recursive self-improvement (RSI). "Directing the overwhelming majority of our compute toward serving people, rather than racing toward recursive self-improvement, is one of the best ways to ensure we develop this technology safely," Zuckerberg said. The concern is that RSI could allow AI systems to rapidly enhance their own capabilities, potentially outpacing human oversight and increasing the risk of losing control over powerful models. Wang similarly described the pursuit of RSI as "one of the most dangerous paths that could lead to losing control of powerful models."

The AI pace debate polarizes the tech industry

The dispute over the tempo of AI development has split the technology sector along distinct lines. Anthropic CEO Dario Amodei has called for a slowdown in frontier AI development so safety mechanisms can keep pace, while OpenAI CEO Sam Altman, Microsoft CEO Satya Nadella, and SpaceX's Elon Musk support a more careful, safety-focused approach without halting development altogether. On the other hand, former President Donald Trump opposes slowing down the US AI race due to concerns about China taking the lead, and David Sacks, a former Trump AI advisor, has said companies should manage their own risks rather than seek government restrictions — while also questioning Anthropic's safety warnings. Investor Michael Burry has raised doubts about whether some calls for a slowdown are partly self-interested, given that they could protect existing AI companies from competitive pressure.