In its recent article, IT Supply Chain, highlights the hidden risks of agentic AI, warning that many organisations are adopting complex systems without fully understanding how they work. As AI agents increasingly communicate in natural language, ambiguity and compounding errors can reduce reliability, often producing confident but inaccurate outputs. Kit Ruparel, Recordsure’s CTO emphasises the need for clear architecture, strong governance, and the ability for systems to acknowledge uncertainty. Ultimately, agentic AI is only as safe as its design making responsible implementation critical.
