Anthropic’s Claude Mythos Now Scanning Code for 150+ Critical Infrastructure Orgs Across 15 Countries

Anthropic is dramatically expanding Project Glasswing, its initiative to hunt down critical software vulnerabilities using AI. Starting Tuesday, roughly 150 new organizations across more than 15 countries gain access to Claude Mythos, the company’s most advanced model for detecting zero-day flaws. The move comes just days after Anthropic filed confidentially for an IPO, following a $65 billion funding round that valued the firm near $1 trillion. Earlier this month, 50 initial partners—including the U.S. government—began using Mythos Preview to scan their codebases. Now, that list extends to power grids, water systems, healthcare networks, telecoms, and hardware manufacturers. “A successful attack on their codebase could be catastrophic,” Anthropic noted in a blog post, estimating that for most partners, a major breach could affect over 100 million people. The expanded cohort includes companies like Okta, Samsung, SK Hynix, SK Telecom, NATO, and the EU’s cyber agency ENISA. Participating countries span Australia, Canada, France, Germany, India, Japan, South Korea, and others. Anthropic expects rivals to soon match Mythos’s capabilities—OpenAI already released GPT-5.5-Cyber to a large testing group. That urgency is driving Anthropic to lock in safeguards now, before the security landscape shifts again.