Unpacking the White House executive order on frontier AI, cybersecurity

After scrapping an initial order anticipated last month, U.S. President Donald Trump issued a much-anticipated executive order addressing cybersecurity concerns related to frontier artificial intelligence development 2 June. 

The signed executive order virtually mirrors the initially anticipated order, with one significant exception. It reduces the timeframe for AI developers to voluntarily submit their models for government review from 90 days in the unsigned prior draft order, down to 30 days.

White House spokeswoman Liz Huston told The New York Times the executive order represented a "common-sense approach of collaborating with industry to balance innovation and security, cementing America’s continued global dominance in AI and cybersecurity."

The order directs key government agencies, including the Department of Treasury, Department of Defense and the National Security Agency, to "develop and maintain a classified benchmarking process to assess the advanced cyber capabilities of AI models and determine the threshold at which an AI model should be designated a 'covered frontier model.'"

Under the order, AI developers would participate in a voluntary framework established by the agencies and would engage with the government to "determine whether model(s) under development meet the designation of 'covered frontier model,'" while providing agencies "access to covered frontier models, subject to appropriate confidentiality, cybersecurity, insider-risk, and intellectual-property protection, use, and nondisclosure requirements, for a period of up to 30 days before they plan to release such models to other trusted partners."

The order also seeks to alleviate industry concerns that it could be used as a mandatory government model-vetting process and states that it should not be "construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models."

Additionally, the order establishes an entity within Treasury Department that would help coordinate patching efforts for software vulnerabilities. It also directs key government agencies to develop action plans to respond to frontier AI cyber risks within 30 days. 

According to Politico, Trump intended to sign the 90-day executive order 21 May, but "abruptly rejected" it after speaking with former White House AI czar David Sacks, who left the administration in March to serve as co-chair of the President's Council of Advisers on Science and Technology. 

The scrapping of the earlier order came amid another Politico report, which cited anonymous White House sources who said key members of the Trump administration fell into three main camps over how to craft AI policy. 

One faction includes Sacks, who favors a pro-innovation approach. Another, led by Secretary of Defense Pete Hegseth, is lobbying the president for tighter restrictions on frontier models, such as Claude Mythos, due to concerns they could be weaponized by foreign adversaries. 

The third faction reportedly comprised White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent, who lobbied for a regulatory framework largely established in the new executive order where companies voluntarily provide the U.S. government with a first look at new models.

In a post on the social platform X, AnthropicAI said the executive order "is an important step in strengthening America's leadership in AI," while OpenAI CEO Sam Altman wrote, the U.S. should "lead on AI by continuing to develop the very best models, making sure they're safe, and getting cyber tools into the hands of trusted defenders." The executive order "gets the balance right," he said.