Apple, Google, and Microsoft join Anthropic's Project Glasswing to defend world's most critical software
12 tech rivals deploy an unreleased AI model and $104M to find vulnerabilities before adversaries do.
In a move signaling an existential threat to global digital infrastructure, 12 of the world's largest tech and finance rivals—including Anthropic, Apple, Google, Microsoft, Amazon, and JPMorgan Chase—have formed Project Glasswing. This unprecedented coalition will deploy Anthropic's unreleased frontier AI model, Claude Mythos Preview, to proactively scan for vulnerabilities in critical software. The partners are committing $4 million in direct funding and an additional $100 million in Anthropic Claude usage credits, indicating a level of cooperation and resource-sharing rarely seen among such fierce competitors.
The initiative is described as a defensive "Manhattan Project" in response to the shrinking timeline between vulnerability discovery and weaponization by adversaries using AI. According to CrowdStrike CTO Elia Zaitsev, the threat window has collapsed from months to minutes. Project Glasswing aims to reverse this trend by using the Mythos model to find thousands of hidden bugs before they can be exploited, securing the foundational software upon which all member companies and the broader internet depend. The project's name, inspired by the resilient and camouflaged glasswing butterfly, reflects its goal of creating transparent and robust defenses.
- Unites 12 rivals including Apple, Google, Microsoft, and JPMorgan Chase in a defensive cybersecurity coalition.
- Deploys Anthropic's unreleased Claude Mythos Preview AI model and $104M in combined funding and credits.
- Aims to find and patch vulnerabilities before adversaries can exploit them, as attack timelines shrink to minutes.
Why It Matters
This collaboration signals a critical shift in defending shared digital infrastructure against AI-powered cyber threats that endanger global stability.