Latest / Cybersecurity Under Pressure. Real Attacks, Real Lessons
Prompt Injection and Agentic AI Security: Executive Brief
Documents and web pages can carry instructions that undermine enterprise AI workflows. This executive brief explores indirect prompt injection, adversarial methods and the move from DevOps to MLSecOps. It connects those technical risks to governance questions for higher-risk AI deployments.
Related CUP episodes:
Why Hidden Text Hacks Enterprise AI
[2026] Urgent: When AI Agents Become the New Insider Threat - Zero Trust
When AI Crossed the Trust Boundary: The OpenAI–Hugging Face Incident
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.