The White House has issued a new executive order establishing a voluntary framework for developers of advanced artificial intelligence models to submit their systems for federal cybersecurity and national security review before public release. The order, titled “Promoting Advanced Artificial Intelligence Innovation and Security,” was signed on 2 June and represents the third major AI directive of the Trump administration’s second term.
Under the new framework, developers of so-called “covered frontier models” may voluntarily grant federal agencies access to their systems for up to 30 days prior to release. Participating developers can collaborate with the government on cybersecurity testing and secure deployment practices during this window.
The order directs the Committee on National Security Systems, Department of War, and Department of Homeland Security to prioritise cyber defence of national security systems and federal networks within 30 days. The Cybersecurity and Infrastructure Security Agency has also been tasked with expanding its cybersecurity services to state and local governments and critical infrastructure operators.
A notable component of the order is the creation of an AI cybersecurity clearinghouse, to be operated jointly by the Treasury Department, National Security Agency, and CISA. The clearinghouse will coordinate vulnerability discovery, software testing, remediation efforts, and vulnerability disclosures in collaboration with the AI industry.
The Attorney General has been directed to prioritise enforcement against criminal actors who use AI systems to unlawfully access computer systems or facilitate cybercrime. Meanwhile, the Office of Personnel Management must expand federal recruitment pathways for cybersecurity specialists within 60 days.
Industry observers note that while participation remains entirely voluntary and the order explicitly rules out mandatory licensing or preclearance requirements, the framework signals a growing expectation that developers of frontier AI systems will engage with national security agencies before release. “This is a notable shift for an administration that previously championed a nearly hands-off approach to AI governance,” said Aria Austin, a technology policy attorney at Holland & Knight.
For AI developers, the immediate practical effect is likely to be increased engagement with cybersecurity agencies, alongside growing pressure from customers and policymakers to participate in voluntary security assessments. The classified benchmarking process that will determine which models qualify as “covered frontier” remains under development, creating some uncertainty about which systems will fall under the framework.