Loading page…
The Microsoft AI chief is challenging the idea that U.S. competition with China requires a hands-off approach, arguing that common safety standards can support innovation rather than slow it.
Listen to this story
The policy divide is shifting toward implementation: Mustafa Suleyman argues that U.S. AI safeguards can establish a shared safety floor without slowing development, while the Trump administration favors voluntary, use-case-specific rules. A June executive order promotes innovation and security but bars mandatory licensing, preclearance, or permitting for new models.
Suleyman made the argument in a CNN interview on Fareed Zakaria GPS, challenging China-based objections to U.S. guardrails.
The June executive order directs agencies toward a voluntary advanced-model framework without mandatory licensing or preclearance.
Michael Kratsios favors safety rules organized by use case and industry rather than one-size-fits-all regulation.
Microsoft AI chief Mustafa Suleyman says the United States should not treat competition with China as a reason to abandon AI safeguards. In a CNN interview, he argued that regulation can build shared safety standards without sacrificing American innovation.
Suleyman made the case on Fareed Zakaria GPS as the White House has emphasized rapid AI development and warned against regulation it considers overly burdensome. His intervention does not call for a development slowdown. Instead, it disputes the premise that safety measures and competitiveness must be opposing choices.
The immediate issue is whether China’s AI progress should change the threshold for U.S. safeguards. Suleyman said concerns about China should not be used as an argument against guardrails, and said policymakers can address safety risks without undermining innovation.
“I don't think we should use China as the bogeyman for not making progress on our own efforts” on AI safety.
Mustafa Suleyman, on CNN’s Fareed Zakaria GPS
Suleyman’s proposed mechanism is regulation as a way to create shared norms and standards. He said that approach can increase safety and “shouldn’t have to slow us down,” framing rules as an operating baseline for developers rather than a brake on the industry.
That position differs from President Donald Trump’s recent calls for largely unfettered AI development. Trump has described fears of catastrophic consequences from uncontrolled AI as a hoax, while arguing that the United States must maintain its lead over China.
The administration’s written policy is not simply a rejection of security concerns. A June executive order says the United States should promote AI innovation and security, work with the private sector, and avoid overly burdensome regulation. It also directs agencies to develop a voluntary framework for certain advanced models and says that framework cannot create mandatory licensing, preclearance, or permitting for new AI models.
Michael Kratsios, Trump’s science and technology adviser, has described a different regulatory philosophy: AI safety should be handled by use case and industry, rather than through one-size-fits-all rules. That leaves a substantive gap with Suleyman’s appeal for shared norms. Neither position, as presented here, identifies which particular safeguards should apply across the AI industry or who would enforce them.
Suleyman’s remarks sharpen the disagreement without resolving it. His claim is that the U.S. can remain competitive while setting a common safety floor; the administration has signaled a preference for voluntary, targeted approaches that avoid broad constraints. The practical next move is whether either side turns that high-level choice into specific standards for AI developers.
Loading discussion...
Join the conversation
Share the tradeoff you think policymakers should prioritize.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.