AI Security Threats: A Practical Guide to Managing Generative AI Risk

This report outlines the landscape of AI security threats—including prompt injection, data leakage in generative AI, model theft, adversarial attacks, risky AI agents, supply chain and third-party risks, AI-powered phishing/quishing, deepfakes, and how generative AI accelerates social engineering—and provides a five-step risk-management framework (inventory, classification, secure deployment, user protection, and resilience), recommended controls (least privilege, logging, vendor due diligence, approval workflows, backups), and vendor evaluation questions, while noting Hornetsecurity’s 365 Total Protection as an example mitigation offering.