Hacker Inti De Ceukelaire: “People used to be the weakest link. Now it’s AI.”

Artificial intelligence promises retailers a revolution in efficiency and customer service. But its vulnerabilities are also becoming increasingly apparent. Ethical hacker Inti De Ceukelaire: “You have to know how a dog behaves before you let it around your children.”

“Way more cyberattacks ahead”

First Ahold Delhaize, then rogue French refrigerators, and more cyber panic at H&M: last week felt like “cyberhack week.” “It’s going to be ‘hack week’ a lot more often,” chuckles ethical hacker Inti De Ceukelaire. Why do hackers target retailers so much? “There’s money to be made there: they’re highly visible and hold large amounts of customer data and payment information.”

The weakest link in corporate cybersecurity? For years, that was the human factor. But according to Inti De Ceukelaire, that vulnerability is now shifting toward artificial intelligence. With the rise of AI agents and automated customer service systems, new opportunities are emerging for both hackers and consumers who want to take advantage of the technology.

De Ceukelaire has already succeeded in manipulating chatbots to gain access to internal files, accounts, and private data. Commercial applications also appear to be vulnerable. He managed to get sales managers to promise discounts via AI systems that they might never have wanted to give. And what happens when a chatbot promises a customer a discount in writing? Who is responsible for that?

Is AI susceptible to advertising?

The relationship between retailers and consumers is also changing. As customers increasingly turn to AI assistants to search for products, compare prices, and make purchases, online stores may see fewer direct visits. This presents a fascinating challenge for marketers. Advertising ultimately revolves around influencing behavior. But what if it’s no longer a human, but an AI agent, making the purchasing decision? Can that computer also be persuaded using advertising techniques?

The ethical hacker explores this question from a psychological perspective. He tests how classic manipulation techniques work on language models. These models sometimes prove surprisingly predictable and susceptible to influence, but they don’t function like the human brain. It’s precisely this last point that we tend to forget (all too) easily.

“Not everything has to be AI”

De Ceukelaire himself uses AI extensively and sees enormous opportunities for businesses. Yet he is frustrated by the speed with which organizations embrace applications without fully understanding how they work. Retailers are automating their customer service and experimenting with autonomous AI agents, often out of fear of falling behind the competition. But not every application actually yields efficiency gains. “Not everything has to be AI,” he emphasizes.

Moreover, human oversight is at risk of disappearing. The “human-in-the-loop” principle—where a human monitors the process and can intervene—is being eroded, he says. “What if an AI makes a mistake? Who’s going to pay for that mistake?” Consider an AI agent that orders a hundred products but specifies the wrong color. Who bears the responsibility when the customer and the online store each have a different order in mind? According to De Ceukelaire, such disputes could create an entirely new set of problems.

Between hype and doomsday thinking

At the RetailDetail Night on November 19, De Ceukelaire will demonstrate how easily AI systems can be manipulated. Can you convince an AI agent to waive a parking ticket? How do you negotiate the price of a car with a computer? And how far can you let a chatbot deviate from its original instructions?

With his interactive show, he aims to debunk both the excessive enthusiasm and the doomsday thinking surrounding AI. The technology can bring retailers significant benefits, but it can also cause considerable harm if they trust it blindly. “It’s like bringing a dog into your home: it can be man’s best friend, but before you let it around small children, you want to know how the animal will react.”

His key message? Companies don’t need to avoid AI, but they do need to understand exactly what they’re bringing into their organizations. Especially when they allow systems to negotiate, sell, and make decisions independently. Because anyone who gives AI responsibility must also know how easily that technology can be misled.

Want to know how secure your AI applications really are? Find out at RetailDetail Night at Trademart Brussels and reserve your ticket now.