Press
Amsterdam, 12 Apr 2026. In this article published on Businesswise, a string of high-profile AI agent failures at companies including Meta and Amazon — leaking data, wiping databases, costing millions — serves as a warning for organisations rushing to deploy autonomous AI. Experts including Rewire's Simon Koolstra argue the problem is rarely the technology itself, but the absence of governance: unclear boundaries, unrestricted access to systems, and no escalation paths to humans. The prescription is straightforward — define what agents may and may not do, limit autonomous action to low-stakes decisions, and ensure explainability before granting autonomy.
