Overview
- Anthropic announced Project Glasswing on Tuesday, giving AWS, Apple, Google, Microsoft, NVIDIA, CrowdStrike, Palo Alto Networks, Cisco, Broadcom, JPMorgan Chase and the Linux Foundation private use of Claude Mythos Preview for defensive work.
- The company says the model has already flagged thousands of high‑severity flaws across every major operating system and web browser and can craft working exploits with little or no human input.
- Anthropic is withholding a general release, offering up to $100 million in usage credits and $4 million for open‑source security groups to help partners find and patch code at scale.
- Examples include a 27‑year OpenBSD bug, a 16‑year FFmpeg flaw, and chained Linux kernel issues, with company benchmarks showing large gains over Opus 4.6 on CyberGym and SWE‑bench Verified.
- The launch follows a late‑March leak that revealed the project, and Anthropic says it is briefing senior U.S. officials as a court fight over a federal move to cut off contracts continues under a judge’s pause.