President Trump's executive order establishes a 30-day voluntary review process for frontier AI models, led by the NSA director, to assess national security risks before public release. This marks a deliberate departure from broader regulatory impulses, prioritizing speed in the US lead over China. The policy directly addresses concerns raised by Anthropic's Claude Mythos deployment, which exposed critical infrastructure vulnerabilities during the firm's Pentagon contract dispute, prompting an emergency Wall Street briefing by Treasury and Fed officials. Unlike Biden-era frameworks that leaned toward mandatory reporting, this approach invites participation from labs like OpenAI and Google while granting the NSA wide discretion on model selection and trusted partner access—a structure that echoes selective intelligence-sharing precedents from the semiconductor export controls era. Mainstream reporting underplays how the shortened timeline and voluntary design mitigate risks of regulatory capture seen in EU AI Act debates, yet it leaves gaps in enforcement consistency that could be exploited in future administrations. Synthesizing patterns from CSIS reports on AI supply chain risks and a 2024 RAND analysis of cyber-AI convergence, the order positions the US to integrate private-sector model insights into critical infrastructure defense without ceding ground in the geopolitical race. The vagueness around 'advanced capabilities' thresholds, however, risks selective application against firms in legal friction with the administration, potentially chilling the very innovation edge Trump seeks to protect.