On April 7th, Anthropic forged a strategic alliance with 12 leading tech companies, including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks, to launch Project Glasswing. As part of this initiative, they unveiled the groundbreaking AI security model, Claude Mythos Preview. This model is designed to bolster cybersecurity defenses and has already pinpointed thousands of high-risk vulnerabilities across popular operating systems and browsers. The overarching goal of Project Glasswing is to harness AI technology to proactively detect, analyze, and mitigate deeply entrenched security threats in critical software, thereby thwarting potential malicious exploitation of high-risk vulnerabilities. Anthropic has pledged up to $100 million in model usage credits, along with a dedicated $4 million donation, to bolster the security of the open-source ecosystem. Eligible organizations can access the model via mainstream platforms, with over 40 critical infrastructure operators already enjoying priority access in the initial rollout phase.