#AI is getting smarter — and riskier. @F5’s Liam Crilly breaks down how agentic AI models open the door to new threats like chain-of-thought injection and data poisoning. 🔒 Learn how to secure inference-time AI:NGINX @nginx · 2025-05-14