"deep, proactive" relationships with AI companies, the memo demands — a simple phrase that frames a broader push by the White House to speed the national-security community’s adoption of frontier artificial intelligence while hardening those systems against theft and manipulation.
President signs national-security memorandum to accelerate AI adoption
President Donald Trump signed a National Security Presidential Memorandum directing U.S. security agencies to move faster in using advanced AI across the military and intelligence community. The memo tasks offices such as the FBI, the Office of the Director of National Intelligence (ODNI), and the Office of the National Cyber Director with identifying where AI can improve government operations, including intelligence analysis and cyber threat detection, while explicitly instructing that the tools cannot be used for unlawful surveillance of Americans.
FBI, ODNI and the Office of the National Cyber Director: build industry ties
The memorandum instructs those agencies to build the "deep, proactive" relationships with AI companies so cutting-edge models can be made available to national security personnel faster. The document reflects a White House view that U.S. security agencies are moving too slowly to adopt frontier AI tools and seeks closer cooperation with developers to accelerate access and integration.
Defense Department and NSA: protect models from theft and compromise
The memo directs senior officials, including Defense Secretary Pete Hegseth and NSA Director Gen. Joshua Rudd, to work with private-sector companies on security protocols meant to prevent advanced models from being stolen, copied or compromised. One technical risk called out explicitly is model distillation — a technique in which an AI system repeatedly queries another to mimic its performance and build a separate model. The White House in April accused China of carrying out "industrial-scale" distillation attacks on U.S. AI systems.
Infrastructure risks: data centers and emergent cyber-focused models
Officials are also ordered to work with industry to secure the infrastructure that supports frontier AI, including the data centers that store the enormous computing power needed to run advanced models. The memo notes data centers have become more attractive targets during periods of geopolitical tension. The administration’s concern has been sharpened by a new class of cyber-focused models — notably Anthropic’s Mythos and OpenAI’s GPT-5.5-Cyber — that can identify vulnerabilities across networks and have driven government discussions over how advanced AI could reshape defensive and offensive cyber operations.
What this means for technologists, policymakers, and critical-infrastructure operators
- Technologists and security teams: Developers are being asked into closer operational relationships with national-security agencies and to collaborate on protocols that guard models and supporting infrastructure. The memo sits alongside an AI security executive order that encourages developers to submit powerful new models to a 30-day government review before public release.
- Policymakers and cyber authorities: More prescriptive guidance is forthcoming — Nick Andersen, CISA’s acting director, said the cyber agency is preparing a binding operational directive focused on AI-enabled cyber threats, signaling a move from voluntary measures toward enforceable requirements in some areas.
- Critical-infrastructure operators: Anthropic recently said it is expanding Project Glasswing — its controlled-access program for giving trusted organizations early access to Mythos — to about 150 additional entities across more than 15 countries, including organizations in water, healthcare, communications and other critical infrastructure sectors; those operators will be both early beneficiaries of capabilities and targets for tightened security collaboration.
The memorandum arrives alongside an administration framework that has already favored voluntary industry cooperation: a recent AI security executive order encourages voluntary submission of models for a 30-day review. Yet the presidential memorandum adds direction for agency-level partnerships and explicit tasks for defense and intelligence leaders to work with companies on protection protocols. With Anthropic expanding Project Glasswing and OpenAI releasing GPT-5.5-Cyber — both developments the memo cites as drivers of concern — the coming weeks should clarify how quickly agencies translate the directive into operational agreements, security controls, and any binding directives like the one CISA is preparing.
