The Open-Source Paradox: Navigating the New Frontier of AI Supply Chain Risk

This blog highlights risks from using public AI model repositories and the emergence of "Shadow AI" and agentic systems, arguing that unvetted models can introduce prompt injection, data poisoning, and other systemic risks; it recommends human-centered vetting, continuous human oversight, workforce training, and the EEG framework (Embrace, Educate, Govern) to manage AI agent risk.