A new study has found that more than half of British companies have faced at least one breach of business policy caused by an AI agent. According to the research, 53% of British Chief Information Officers (CIOs) reported such incidents, a figure much higher than the global average of 31% for rogue AI actions. The study, conducted by Harris Poll on behalf of the data science company Dataiku, surveyed 685 CIOs from the UK, France, Germany, the US, UAE, Japan, South Korea, and Singapore. The findings point to challenges in managing AI within organizations, especially in the UK. Here, 87% of British CIOs described "AI-Agent sprawl" as a major problem. This term refers to the fast-paced introduction of custom AI tools in the workplace, often without proper oversight. The study found that only 49% of British CIOs could provide a clear record of an AI agent's actions, and 68% believed it would take more than 24 hours to detect and contain the effects of a rogue AI agent. CIOs are increasingly under pressure to demonstrate the value of AI investments. Many fear budget cuts or job losses if they fail to meet performance goals. Faye Murray, Field Chief Data Officer at Dataiku, said the high number of reported AI policy violations in the UK shows that simply monitoring AI is not enough. She stressed that accountability should be built into AI systems from the beginning, rather than added later after an incident occurs. Florian Douetteau, CEO of Dataiku, said the industry has focused too much on the independence of AI systems and not enough on their management. He explained that monitoring tells CIOs whether an AI agent is running, but management ensures it has the right to keep running. Currently, few companies can effectively stop a rogue AI agent, often resorting to shutting it down without a clear plan.