Cloudflare AI Security for Apps Reaches GA — Free AI Discovery for All Plans, IBM and Wiz Partnerships for Agent Threat Mitigation

Cloudflare announced the general availability of its AI Security for Apps product on March 11, 2026, creating the first major CDN/WAF-native security layer specifically designed for AI-powered applications and autonomous agents.

PRODUCT CAPABILITIES:

1. AI ENDPOINT DISCOVERY (NOW FREE): Automatically identifies LLM-powered endpoints across web properties regardless of hosting provider or model. Starting immediately, this is free for ALL Cloudflare plans including Free, Pro, and Business — giving every organization visibility into where AI is deployed across their internet-facing apps, including shadow AI deployments.

2. THREAT DETECTION: Detects malicious or off-policy behavior targeting AI endpoints, including prompt injection attempts, sensitive information disclosure, and unbounded consumption attacks as cataloged in the OWASP Top 10 for LLM Applications.

3. MITIGATION: Threats are mitigated via the familiar Cloudflare WAF rule builder, allowing security teams to apply the same workflow they use for traditional web security to AI-specific threats.

4. CUSTOM TOPIC DETECTION: New capability allowing organizations to define custom content categories for their specific AI use cases.

STRATEGIC PARTNERSHIPS:

• IBM COLLABORATION: IBM has chosen Cloudflare to deliver AI security to its cloud customers, representing a significant enterprise channel partnership
• WIZ INTEGRATION: Partnership with Wiz (recently acquired by Google) to provide mutual customers with a unified view of their AI security posture across cloud and application layers

SECURITY LANDSCAPE CONTEXT:

The launch comes as AI agent security reaches crisis point:

• 48% of cybersecurity professionals consider agentic AI the top attack vector for 2026 (Dark Reading poll)
• NIST published formal RFI on securing AI agent systems in January 2026
• OWASP released Top 10 for Agentic Applications in December 2025
• Cisco State of AI Security 2026 found 83% of businesses plan agentic AI but only 29% feel ready to secure it
• CVE-2026-25253 exposed 17,500+ OpenClaw instances, Bitdefender found 824+ malicious skills

Cloudflare sits at the reverse proxy layer, protecting AI applications regardless of model, framework, or hosting provider — making security infrastructure-level rather than application-level.