“Too Dangerous to Release” — Or Just Too Expensive? The Real Reason Anthropic Is Hiding Its Most Powerful AI
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Anthropic launched Claude Mythos Preview in April 2026 through a gated program called Project Glasswing, restricting access to ~40 vetted organizations. The official rationale is that the model can autonomously discover and exploit zero-day vulnerabilities at scale, making broad release dangerous. However, circumstantial evidence — including simultaneous compute procurement deals, capacity admissions, hiring patterns, and a leaked internal draft describing Mythos as 'expensive to run' — suggests compute constraints are also a significant factor. The most honest assessment is that both security concerns and infrastructure economics are real, mutually reinforcing reasons for the restricted rollout, but Anthropic has strong incentives to publicly emphasize only the more flattering safety narrative.
