Latest / Cybersecurity Under Pressure. Real Attacks, Real Lessons
Indirect Prompt Injection: How Hidden Text Targets Enterprise AI
Indirect prompt injection can hide instructions in material an enterprise AI system is expected to read. Explore how agentic AI expands the consequences of that trust problem. The episode examines security decisions around models that can interact with systems and perform actions.
Related CUP episodes:
Prompt Injection and Agentic AI Security: Executive Brief
Agentic AI Security: Controlling Autonomous System Access
AI Sandbox Security: The OpenAI–Hugging Face Trust Boundary
Kinetic AI and Critical Infrastructure: Connecting Cyber and Physical Risk
Siemens S7 PLC Security: How AI Changes…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.