Anthropic holds Claude Mythos in lab and stands up Project Glasswing — $100M credits to AWS, Apple, Google, Microsoft, NVIDIA, JPMorgan, Linux Foundation

Anthropic confirmed Claude Mythos Preview will not be publicly released. Instead, the model is deployed through Project Glasswing — a consortium of AWS, Apple, Cisco, CrowdStrike, Google, JPMorgan Chase, Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. Anthropic is committing $100M in usage credits. Glasswing partners will use Mythos to identify and patch vulnerabilities in critical software before the model's capabilities reach adversarial hands.

The structural move is to convert a capability-holding decision into an infrastructure-protection program. Mythos autonomously discovered and exploited zero-day vulnerabilities in every major OS and web browser during internal testing. The traditional disclose-and-ship pathway would have produced an immediate adversarial-use risk; Glasswing instead routes the capability into defensive hands first.

For the safety-first posture debate, this is the cleanest example to date of a frontier lab translating a capability-hold into a structured industry program rather than a press release. Compared to the pattern Anthropic's $1.25B/month compute deal implied — that the lab needed to keep shipping to support the capital commitment — Glasswing demonstrates a third path: ship the capability to a controlled set of customers under contractual safety obligations. See our analysis →.

Anthropic — Claude Mythos Preview → · InfoQ — Anthropic Claude Mythos → · Just Security — Anthropic Mythos analysis →