Anthropic expands Claude Mythos / Project Glasswing to critical infrastructure partners

“Anthropic's AI is so good at breaking software they won't fully release it, but they're putting it into critical infrastructure defense.”

8.0Weirdness

Why It Matters

Frontier models are now powerful enough at vulnerability discovery that access is gated, then routed into critical infrastructure defense.

Evidence

TechCrunch; Anthropic Project Glasswing; Anthropic initial update.

Caveat

More builder/security than mass-culture; strongest as a smart second segment or newsletter/video essay, not necessarily the lead.

Signal Read

Needs scoring pass

Source Trail

Daily scan: 2026-06-02