Trump Orders AI Vetting as New Zealand Gains Mythos Access

President Donald Trump signed an executive order requiring the vetting of advanced AI for national security risks. The order seeks to harden information systems and requests that AI companies voluntarily provide the U.S. government access to covered frontier models for review up to 30 days before their general release.

In a separate development, New Zealand's National Cyber Security Centre gained access to Mythos AI, a frontier model developed by Anthropic known for significant hacking capabilities. This access is part of Project Glasswing, an industry collaboration designed to test potentially vulnerable products defensively before they are released to the public.

Deputy director-general for cyber security Catriona Robinson confirmed the model will be used under a cyber defense mandate to better understand the evolving threat landscape. Anthropic has distributed the Mythos model to 150 organizations worldwide for similar cybersecurity testing to strengthen national security missions.