Cloudflare wants to put a firewall in front of your LLM

Cloudflare announced “Firewall for AI,” an enterprise WAF enhancement for LLM applications that introduces Advanced Rate Limiting to mitigate abusive traffic and Sensitive Data Detection to prevent leakage of secrets and financial data in model outputs. The company plans a beta prompt validation feature to score and categorize prompts for injection risk and enable rules-based blocking (including sensitive topics), and it can be deployed in front of any model proxied through Cloudflare, reflecting broader industry moves to strengthen AI security.