This session will explore the specific ways AI poses new risks to enterprise security, from deepfakes and scalable phishing attacks, to data poisoning and prompt injection, and beyond. It will cover both current threats and hypothetical but plausible attack scenarios that security teams need to understand. Attendees will learn how attackers exploit the unique properties of AI models, and why traditional security tools may not be enough. Ideal speakers will be able to connect deep technical knowledge with real-world implications. Potential topics include:

• AI-based attack automation
• Prompt injection and jailbreak techniques
• Model inversion and training data leakage
• Adversarial examples and AI model manipulation