Trump plan to test AI models has a problem—US security teams were gutted by DOGE

President Donald Trump signed an executive order Tuesday establishing a voluntary federal safety-testing regime for frontier artificial intelligence models, but critics warn the framework lacks enforcement power and may not prevent dangerous deployments [1]. The order, titled “Promoting Advanced Artificial Intelligence Innovation and Security,” directs the National Security Agency to create a classified benchmarking process to define which systems qualify as “covered frontier models” with advanced cyber capabilities [3]. Developers can then submit those models for government review up to 30 days before wider release to trusted partners, a window shortened from the 90 days proposed in an earlier draft [2][4]. The order explicitly bars any mandatory licensing, preclearance, or permitting, and no AI firm is required to participate [3]. Trump’s directive promises not “to stifle this innovation with overly burdensome regulation” [1]. Trump scrapped a planned signing event last month after several AI CEOs declined to attend on short notice, citing concerns the earlier draft had become a “blocker” on innovation [2]. The version he signed differs from the leaked draft mainly in shortening the government’s testing access window, a concession reportedly driven by White House worries about slowing US firms in the AI race [4]. In the order, Trump wrote, “we will continue to lead an America First cybersecurity effort that enhances both our national security and our global AI dominance” [1]. The order assigns the NSA, the US Treasury Department, and the Cybersecurity and Infrastructure Security Agency to build a “cybersecurity clearinghouse” to scan and patch vulnerabilities at scale [2]. Yet CISA was one of the hardest-hit agencies during Department of Government Efficiency cuts in 2025, with top officers fired, cybersecurity contracts canceled, and institutional knowledge depleted [4]. Matthew Ferren, an international affairs fellow at the Council on Foreign Relations, described the order as “best understood as an attempt to engineer a cybersecurity window of opportunity” — granting US defenders preferential early access to new model capabilities while attempting to delay adversary exploitation [4]. The plan depends on a federal cyber workforce that does not currently exist at the needed scale [4]. Agencies have 30 days to prioritize cyber defenses for national security systems and establish the AI cybersecurity clearinghouse, and 60 days to design the voluntary submission process and expand hiring pathways for cybersecurity specialists [3]. The Office of Management and Budget is directed to hunt across existing federal grant programs for funding that could be steered toward advanced AI vulnerability detection, a sign that no new appropriations are attached [4]. The order arrives in response to public concern over the cybersecurity risks posed by Anthropic’s model Mythos [2]. For AI firms, the practical impact is small: submission is voluntary, the 30-day window is tight enough that most labs can structure release schedules around it, and the EO contains no enforcement mechanism against companies that decline to participate [4].