A report reveals 54% of enterprises have experienced AI agent security incidents, with many still allowing shared credentials. The study highlights an 'agent security gap' where autonomous agents proliferate faster than necessary controls.