RAG can make AI models riskier and less reliable, new research shows

According to Bloomberg, the increasingly popular AI framework can vastly increase your chances of getting dangerous answers. What can you do?