Cloudflare Launches Identity-Aware AI Gateway to Catch Rogue Agents and Overspending
Cloudflare Blog · rss · 2026-08-05
Cloudflare announced two major updates to its AI Gateway to help enterprises track and control internal AI usage.
- Identity-aware AI Gateway (Open Beta): Integrating with Cloudflare Access, it authenticates requests via IdPs like Okta. Every API call is tied to a verified user identity, enabling per-user budget caps or automatic fallback to cheaper models when limits are hit.
- User Insights (GA): Establishes a behavioral baseline for every account using existing traffic. Instead of just tracking absolute spend, it analyzes sessions against a user's 30-day p95 cost. If a session exceeds 2x the user's baseline and falls into the top 1% (p99) of company-wide spend, it triggers an anomaly alert.
This relative deviation approach effectively distinguishes a busy engineer from a rogue agent, filtering out noise from low-cost micro-spikes.
More from Safety
- MCP Beyond APIs: The New Security Contract Introduced by Agents — drhyrum · 2026-08-05
- Ex-Anthropic Researcher Warns Small Models Bypassed Controls to Attack Real Bank During RL — gerardsans · 2026-08-05
- Malicious GitHub Repo Disguised as Crypto Exploit Exposed via LLM-Assisted Review — RSync25 · 2026-08-05
- ByteDance Bans Employees from Distilling US Frontier Models to Protect TikTok — ns123abc · 2026-08-05
- White House Exempts Open AI Models from New Frontier Capability Testing Framework — ylecun · 2026-08-05
- Meta's Smart Glasses Spark EU Privacy Fears and 20k GDPR Complaints — nordicinst · 2026-08-05