NCCapital

Meta CEO Mark Zuckerberg has entered the growing debate over artificial intelligence safety, arguing that AI companies already have strong incentives to make their systems safe without collectively slowing the pace of development.

Zuckerberg outlined his position in a social media post Tuesday, pushing back against calls from Anthropic CEO Dario Amodei and other AI leaders for companies to coordinate on slowing the development of increasingly advanced models.

According to Zuckerberg, competition, user expectations and potential legal liability give AI laboratories a “natural incentive” to prioritize safety. He argued that companies can independently determine when they need to slow or delay their own work rather than requiring an industrywide pause.

“Every lab has the responsibility and incentive to move at the pace required to train its models safely,” Zuckerberg wrote, emphasizing that individual companies have the ability to take action when safety concerns arise.

Meta has pointed to its own development decisions as an example. Zuckerberg said the company delayed the release of its Muse AI agent for several months while it focused on safety and security, rather than asking competing companies to delay their own products.

The comments come after Amodei published a lengthy argument calling for a slower pace of development for the most advanced AI systems. OpenAI CEO Sam Altman and SpaceXAI CEO Elon Musk have also expressed support for aspects of Amodei's approach, creating an unusual public discussion among leaders of competing AI companies.

Amodei has argued that increasingly capable AI systems could create significant risks if their capabilities advance faster than researchers' ability to understand and control them. His proposals include greater use of independent safety evaluators and other forms of external oversight.

Zuckerberg does support that particular element of the safety discussion. He said engaging independent evaluators and advisers is an industry best practice and suggested that a larger and more diverse ecosystem of evaluators could improve the process.

The disagreement therefore centers less on whether AI safety matters and more on how companies should respond to the risks. Zuckerberg favors companies taking individual responsibility for safety, while Amodei and other executives have called for greater coordination and a slower pace of frontier-model development.

The debate comes as AI companies race to develop increasingly autonomous systems. Zuckerberg specifically emphasized the importance of “alignment,” the effort to ensure that AI systems behave in ways that match user intentions and expectations.

The split among technology executives highlights the broader uncertainty surrounding how quickly advanced AI should develop and what safeguards should accompany that progress. Companies are simultaneously investing heavily in more capable systems while developing new testing, evaluation and security practices.

For Meta, the debate also comes as the company continues expanding its AI ambitions. Zuckerberg's comments indicate that Meta intends to continue developing advanced systems while using internal delays and external evaluation to address safety concerns rather than pursuing an industrywide slowdown.

As AI capabilities continue advancing, the disagreement between advocates of coordinated limits and companies emphasizing individual responsibility is likely to remain a central issue for the technology industry.