Why UK Banks Could Soon Be Banned From Using 'Dangerous' AI

Jan de Vries · 2026-08-13

Artificial intelligence is everywhere these days. It writes our emails, powers our search results, and even helps us decide what to watch on a lazy Sunday. But when it comes to the world of high finance, AI is starting to look less like a helpful assistant and more like a wild card that could burn the whole house down. A new study out of Durham University has put a spotlight on something that's been simmering in the background for a while now: the fact that banks are using AI in ways that could be genuinely harmful to consumers. The researchers aren't just raising eyebrows. They're calling for outright bans on certain types of AI that they consider too dangerous to deploy in financial services. This isn't about your bank's chatbot getting a little confused about your account balance. This is about algorithms making life-altering decisions with zero human oversight. And if you think that sounds dramatic, you might want to keep reading. ### The Core Problem: AI That Makes Decisions Without a Safety Net The Durham University research points to a fundamental issue: many banks are using AI systems that operate as black boxes. You feed them data, they spit out a decision, and nobody can really explain how they got there. That's a terrifying thought when the decision involves whether you get approved for a mortgage, a car loan, or even a basic checking account. Think about it like this. If a human loan officer denies your application, they can tell you why. Maybe your credit score is too low, or your debt-to-income ratio is out of whack. But if an AI denies you, the reasoning can be completely opaque. It might have learned patterns from historical data that are biased, discriminatory, or just flat-out wrong. The researchers are specifically concerned about AI systems that make autonomous decisions without any meaningful human review process. They argue that when these systems fail, they fail on a massive scale. A bank teller might make one bad call an hour. An AI can make millions of bad calls in a single second. ### Why This Matters for Your Wallet Here's where this gets personal. If you've ever been frustrated by a bank's seemingly arbitrary decision, you've probably already felt the edge of this problem. But the stakes are much higher than just a denied credit card application. Consider what happens when AI is used to detect fraud. These systems are trained to flag suspicious activity. But they're also prone to false positives. Imagine your card getting frozen because you bought a plane ticket and a hotel room in the same day. That's annoying. Now imagine the AI decides your entire account is compromised and locks you out for a week while you're trying to pay your rent. That's a real problem. - AI credit scoring can lock people out of the housing market based on hidden biases - Automated fraud detection can freeze accounts on a whim, leaving customers stranded - Chatbots and virtual assistants can give incorrect financial advice that customers trust blindly - Algorithmic trading systems can trigger flash crashes that wipe out savings These aren't hypothetical scenarios. They're happening right now, every day, at banks around the world. ### What a Ban Would Actually Look Like The word "ban" sounds heavy, and it is. But it's not like the researchers are suggesting we throw every computer in the banking sector into the ocean. They're proposing a targeted approach that would prohibit the use of AI in specific high-stakes scenarios where the risk of harm outweighs the potential benefits. One area that's ripe for regulation is the use of AI in credit decisions. If a bank can't explain why an AI system rejected an applicant, then that system probably shouldn't be making the final call. The same logic applies to insurance pricing, investment advice, and any other area where a machine's decision can have a lasting impact on a person's financial life. The call isn't just coming from academia, either. Consumer advocacy groups have been pushing for stricter AI oversight for years. And regulators in the UK and Europe are starting to listen. The EU's AI Act is already setting the groundwork for a risk-based approach to AI regulation, and this new research adds more fuel to that fire. ### The Banking Industry's Response Predictably, the banking industry isn't thrilled about the idea of bans. Banks argue that AI helps them cut costs, reduce fraud, and serve customers more efficiently. And they're not wrong. AI does all of those things. But the industry's track record on self-regulation is, to put it kindly, mixed. There's also a competitive angle here. If UK banks are forced to abandon certain AI tools while their international rivals are free to use them, that could put them at a disadvantage. But the flip side is that trust is a competitive advantage too. A bank that can say "we never let a machine make a final decision about your money" might actually win over a lot of customers. ### What Happens Next The debate over dangerous AI in banking is only going to get louder. As the technology becomes more sophisticated, the potential for harm grows right along with it. The Durham University research is a wake-up call, but it's also a starting point for a conversation that needs to happen at every level, from bank boardrooms to government regulators. For now, the best advice for consumers is to stay informed. Ask questions when you get a surprising decision from your bank. Request explanations. And don't be afraid to shop around if you feel like you're being treated unfairly by a machine. The future of banking is going to be shaped by AI, whether we like it or not. The question is whether we're going to let it run wild or put some guardrails in place before it does real damage. That's a choice we all have a stake in.