Trump memo pushes national security agencies to move faster on AI

President Donald Trump signed a National Security Presidential Memorandum on Friday aimed at accelerating the adoption of advanced artificial intelligence across U.S. national security agencies. The memo addresses concerns that agencies like the FBI, Office of the Director of National Intelligence, and Office of the National Cyber Director are moving too slowly to integrate frontier AI tools, despite rapid technological advancements and competition from rivals like China. The directive calls for ‘deep, proactive’ partnerships with AI companies to expedite access to cutting-edge models for intelligence analysis and cyber threat detection. It also prohibits unlawful surveillance of Americans, addressing long-standing civil liberties concerns over data collection practices. A key focus is protecting U.S.-developed AI models from foreign adversaries, including China, which the White House accused in April of conducting ‘industrial-scale’ model distillation attacks. The memo tasks senior officials, including Defense Secretary Pete Hegseth and NSA Director Gen. Joshua Rudd, with developing security protocols to prevent theft or compromise of advanced models. The administration also targets securing AI infrastructure, such as data centers that support high-compute models, which have become prime targets during geopolitical tensions. Trump recently signed an AI security executive order encouraging voluntary industry cooperation, including a 30-day government review for powerful new models before public release. The shift reflects growing concerns over AI-enabled cyber threats, including models like Anthropic’s Mythos, which can rapidly identify network vulnerabilities. Last week, Anthropic announced expanding Project Glasswing, a controlled-access program granting early model access to trusted organizations. The memo aligns with broader efforts to mitigate risks while leveraging AI for national security advantages.