New U.S. AI Order Balances Innovation With National Security Oversight
Executive order establishes a voluntary framework for developers of frontier AI systems to work with federal agencies on security assessments before deployment.
PRESIDENT Donald Trump has signed a new executive order establishing a voluntary framework for the U.S. government to review some of the country’s most advanced artificial intelligence models before they are released to the public, marking a significant shift in Washington’s approach to AI governance.
The order, signed in June 2026, seeks to strengthen national security while preserving America’s leadership in artificial intelligence. Rather than imposing mandatory licensing or government approval, it encourages developers of “covered frontier models” to voluntarily provide federal agencies with secure access to their systems for up to 30 days before public release.
Security Focus Rather Than Regulation
According to the White House, the initiative is designed to identify potential cybersecurity risks associated with increasingly powerful AI systems capable of sophisticated cyber operations.
The executive order directs agencies, including the National Security Agency (NSA), the Cybersecurity and Infrastructure Security Agency (CISA), the Department of Homeland Security and the National Institute of Standards and Technology (NIST), to establish classified benchmarking procedures for determining which AI models qualify as frontier systems requiring additional security collaboration.
Importantly, the directive explicitly states that it does not create a mandatory licensing, permitting or government pre-clearance requirement for AI developers before releasing new models. Participation remains voluntary.
Industry Pushback Led to a Narrower Policy
Reports indicate that an earlier draft of the executive order proposed allowing government access to AI systems as much as 90 days before public release. However, following concerns from technology companies that such requirements could slow innovation and weaken America’s competitiveness against China, the administration reduced the proposed review period to 30 days and maintained the voluntary nature of the programme.
Major AI developers have broadly welcomed collaboration on cybersecurity, although some industry experts continue to express concern over how frontier models will be selected and how confidential commercial information will be protected during government evaluations.
Balancing Innovation and National Security
Technology analysts view the executive order as an attempt to strike a middle ground between safeguarding national security and avoiding regulatory burdens that could hinder AI innovation.
Beyond model reviews, the order also directs federal agencies to strengthen cybersecurity across critical infrastructure and prioritises prosecution of crimes involving AI-assisted hacking and cyber intrusions.
The policy is expected to shape future discussions on AI governance as governments worldwide seek ways to manage the opportunities and risks associated with rapidly advancing artificial intelligence technologies.
