The Mythos Race: Trump’s New EO and Glasswing’s Expansion

This blog post reviews three AI-related cybersecurity developments: the June 2, 2026 White House executive order directing federal AI cybersecurity guidance and NSA benchmarking; Anthropic’s expansion of Project Glasswing, which uses the Claude Mythos model to scan partner codebases and reportedly surfaced over 10,000 high/critical security flaws across initial partners; and a discussion of an industry metric (“token value per watt per user”) that may determine competitive advantage in AI. The author frames these items as evidence of rapid change in AI and cyber defense, praises proactive vulnerability scanning while warning about the looming risk of powerful AI tools being weaponized, and emphasizes the need for broader collaboration and faster institutional adaptation.