Defend against frontier cyber models: Cloudflare's architecture as customer zero

Cloudflare published a collection of security-focused posts covering defenses for frontier AI/LLM threats (Project Glasswing findings, architectural guidance for protecting against model-driven attacks), integration of Cloudforce One threat intelligence into real-time WAF rules (cf.intel fields), and GA of several security products (AI Security for Apps, post-quantum IPsec via hybrid ML‑KEM, Client‑Side Security, account abuse protections). They describe mitigations and detection improvements (managed OAuth and scoped non‑human identities, automated revocation, Log Explorer enhancements, a new