[un]prompted: Key Insights from the AI Security Practitioners Conference – FireTail Blog

This blog post summarizes key takeaways from the AI Security Practitioners Conference, highlighting accelerated exploit discovery using LLMs, new AI-specific attack surfaces (prompt injection, evasion, RAG pipeline risks), the need for threat modeling tailored to LLMs, and the rise of defense-in-depth tooling and agentic automation for AI security.