AI security is critical. What every developer needs to know. Threats: 1. Prompt Injection: Crafted inputs manipulate AI 2. Data Poisoning: Corrupted training data 3. Model Extraction: Reverse-enginee