Prompt Injection Attacks Pose Top Security Risk for AI

Prompt injection is the primary AI security risk, tricking chatbots into following malicious instructions instead of user commands. OpenAI acknowledges it’s difficult to solve.