Semana warns shadow AI inside firms is a growing security blind spot

Semana’s column puts a Latin American lens on a global issue: AI adoption is no longer something central IT rolls out; it’s something employees quietly bring in. Shadow AI—staff pasting source code or customer data into consumer tools without approval—means that organizations are already running on AI long before they have monitoring, governance, or legal frameworks in place. The piece ties this to recent incidents like Meta’s AI support bot being socially engineered into account takeovers, underscoring that the weak link is not just model vulnerability but how humans route sensitive processes through opaque systems. For the AGI race, this is a reminder that capability deployment is outpacing institutional adaptation everywhere, not just in Silicon Valley or Beijing. As companies in markets like Colombia, Mexico and Brazil lean on off‑the‑shelf AI to stay competitive, they are inadvertently stress‑testing real‑world failure modes and data‑leak pathways at scale. That will create demand for local security vendors, policy responses, and eventually regional AI infrastructure that promises more control than global, US‑centric tools.