AI agents may shop for you in the future. But who pays the bill when things go wrong?

Dubai: A future where artificial intelligence handles your online shopping may not be far away, but it raises a question that touches daily life. Who pays the bill when an AI buys something you never intended to purchase? At the Dubai Future Forum, Omar Al Olama, Minister of State for Artificial Intelligence, Digital Economy and Remote Work Applications, and Michael Miebach, CEO of Mastercard, discussed the growing role of autonomous AI agents, which are increasingly capable of handling tasks on behalf of users.

If children can already make accidental in-app purchases on tablets, what happens when AI agents start making decisions on a user’s behalf? And what happens when a so-called AI agent is not an agent at all, but a fraudulent system designed to mimic one?

Michael Miebach framed this as one of the most urgent consumer questions of the coming years. “In an agentic transaction, is it your kid that should ideally not do this transaction, or is it even you? Or is the agent, the AI enabled agent, actually a real agent, or is it a fraudster that set up a fake agent?” he said. He warned that the problem becomes much more serious when an AI completes transactions independently. “What happens if something goes wrong in the world of an AI generated transaction is what do you do as a consumer? You say, I never intended to do this transaction.”

He stressed the need for safeguards, verification layers and dispute-resolution systems that are designed for a world where AI may act before a human reviews a decision. “We have to build in the safeguards. We have to build in the controls. And that is what our business does for a living. That is what regulators talk about,” he said.

The UAE’s advantage lies in a different mindset

Al Olama noted that the UAE is well-positioned to test many of these technologies because of its distinctive national approach to innovation. While many countries question new ideas with “Why?”, the UAE often begins with “Why not?”. He said that this mindset enables pilots, experimentation and rapid learning in controlled environments.

“There is a way of thinking here that is probably our secret as a country,” he said. “In most countries, when you say you want to do something, the response you get is why. Here, the response is why not. Let us try it, let us run a controlled environment and see how it works.” He added that this approach stems from a belief that the future can always be improved. “Why not try it? Because how are you ever going to get to that next step?”

Miebach agreed, noting that the country’s openness to experimentation allows companies to test ideas that might face resistance elsewhere. “There is no question on behalf of the government here that we should try that,” he said.

Data will define who benefits and who is left behind

Miebach also warned of a growing divide between countries with structured data systems and those without them. “AI, in itself, does not actually do anything. AI will do something in combination with data, structured data,” he said. He explained that countries with strong data foundations can harness AI to improve services and economic outcomes, while others risk falling behind.

“If you have parts of the world where there is not structured data available, then you will get to a potential world of an uneven playing field,” he said. He noted that this could deepen inequalities and prevent emerging markets from benefiting from AI tools that could otherwise improve quality of life.

People will only adopt AI-powered financial tools if they trust the system. A single unintended purchase, a fraudulent agent, or a dispute with no clear resolution process could discourage whole communities from using AI for everyday needs. For that reason, safety and innovation must rise together.