The debate over whether frontier AI development should be paused has intensified in recent days.

On Wednesday, Meta CEO Mark Zuckerberg addressed the controversy publicly for the first time in a post on X, stating that AI labs should incorporate independent evaluation bodies and external consultants, embedding safety reviews into routine development workflows. He described this approach as "industry best practice."

This stance contrasts with the "slowdown" proposition previously put forward by Anthropic CEO Dario Amodei. Over the past weekend, Amodei called for a measured deceleration in the development of the most advanced AI systems to grant researchers additional time to assess potential risks. OpenAI CEO Sam Altman and Elon Musk subsequently expressed support for this view, while Nvidia CEO Jensen Huang emphasized on the same day that AI safety is primarily an engineering issue.

Zuckerberg and Huang Argue Safety Controls Should Be Embedded in Development

Zuckerberg cited Meta's AI tool, Muse, as an example, noting that the company delayed its release for several months due to security and safety concerns. However, he did not ask other companies to slow down in tandem, instead choosing to integrate safety checks into the standard development process.

Huang shares a similar perspective. He stated that safety is the "top priority" in AI development, but does not require new laws or regulations to address it. "We don't need new laws, we don't need new regulations," he said. He believes companies can mitigate risks through testing, evaluation, and safety validation, and that market forces can also serve as a constraint.

Huang also rejected the notion of framing safety and development speed as a simple trade-off. He argued that businesses can advance innovation while controlling the pace of releases; if a product carries clear safety risks, it should be paused and the issues resolved first.

Altman's Position Softens, Markets Focus on Regulatory Impact

Altman previously supported Amodei's call for a measured slowdown of frontier AI development, but his latest remarks take a more moderate tone. He stated he is "very confident" the AI industry can continue to advance technology while avoiding major harms, and confirmed that OpenAI is cooperating with Anthropic and Google DeepMind on AI safety initiatives.

Altman also expressed disagreement with simplifying the current debate to a binary choice of "slow down vs. speed up," emphasizing that individual companies should possess the capability to independently drive safety improvements.

There is currently no major division among AI companies regarding the need to strengthen safety governance. The core contention lies in whether development should be decelerated and whether safety governance should rely more on corporate self-regulation or external oversight.

This divergence is beginning to attract attention from capital markets. Investors are concerned that if regulatory coordination is further strengthened, it could impact the pace of AI model development, with knock-on effects on compute demand, data center construction, and capital expenditures by chip companies. Conversely, if the industry primarily adopts internal governance and third-party assessment pathways, related investment rhythms could remain relatively stable.