Cloudflare tested Anthropic's Mythos Preview LLM on more than fifty internal code repositories as part of Project Glasswing and showed it could build multi-stage exploit chains from vulnerability primitives

— The model generated working proofs of concept in isolated scratch environments.

Original post

OPSS#1992Steven Sinofsky|@STEVESI

Sane and rational view of Mythos and CVEs. https://blog.cloudflare.com/cyber-frontier-models/

7:37 AM · May 19, 2026 View on X

POSTSK#505Séb Krier|@SEBKRIER

Interesting details in Cloudfare's post about Mythos Preview. Harnesses and division of labour is key! https://blog.cloudflare.com/cyber-frontier-models/

9:48 PM · May 19, 2026 · 966 Views