The dark side of AI: Security risks developers ignore until It’s too late

AI is everywhere, and most teams are building with it faster than they can secure it. This talk exposes the real security risks hidden behind modern AI features, from prompt injection and data leakage to model manipulation and insecure integrations. Attendees will learn the mistakes developers and companies keep repeating, why they happen, and how to protect applications before attackers exploit them. If they are shipping AI into production, this is the crash talk they cannot afford to skip.