NSA Leverages White House Directives to Gain Access to AI

The National Security Agency (NSA) is pursuing access to all artificial intelligence (AI) models, as reported by NextGov/FCW. NSA Deputy Director Tim Kosiba stated this at an Intelligence and National Security event in Bethesda, Maryland. The NSA is collaborating with leading AI developers in accordance with White House directives on AI, according to the report.

During the event, Kosiba did not disclose the specific companies, models, or developers the NSA is engaging with. The agency's goal is to scrutinize potential hacking abilities of these models, which could enhance defensive capabilities and strengthen cybersecurity of critical infrastructure.

A June executive order empowers the NSA to determine which AI systems are "covered frontier models" and require developers to provide access for 30 days before the models' public release. Additionally, a June White House directive obliges intelligence and defense agencies to form partnerships in the AI industry, make advanced models available to agency personnel, and avoid over-reliance on a single company.

Google DeepMind, Microsoft, and xAI have previously agreed to share their frontier AI models with a government agency for national security testing before public release. The stakes are rising, as frontier models are increasingly adept at identifying and exploiting software vulnerabilities, which could broaden both defensive capabilities and the attack surface.