Mortgage firms risk becoming hooked on artificial intelligence (AI) tools before finding themselves exposed to rising costs, a roundtable at Mortgage Strategy’s MIT Live 2026 has heard.

One attendee compared the business model behind AI to that of “a heroin salesman”, warning that providers could encourage firms to become reliant on their products before increasing prices.

The participant at the event, which was run under the Chatham House rule, urged mortgage firms to think carefully about the long-term cost of adopting AI, warning against “hooking yourself to a cost you cannot control”.

They added: “My fear is the more you rely on AI, the greater the risk you run into trouble in a few year’s time.”

The comments came during a roundtable discussion on the opportunities and risks of AI automation, as mortgage businesses face growing pressure to use the technology to cut costs, speed up processes and remain competitive.

There are regulatory risks to AI, such as where staff use the technology in an uncontrolled way, such as outside office hours and office systems, the roundtable heard.

One participant said mortgage businesses using AI needed to “control the perimeter” and the exfiltration of data. This could involve locking down down work devices and controlling access to company information and AI, they went on.

Another roundtable participant noted the widespread mortgage market enthusiasm for adopting AI, but warned about the risks of over-extending.

“It’s like a turbo booster in your car,” they said. “It makes you faster, but if you then go too fast your chassis breaks and the wheels fall off.”

The roundtable guests then moved on to debate other areas of the regulatory risk of AI to mortgage businesses.

One attendee noted that in legal terms, AI has ‘no chain of custody’. In other words, it has no practical way of explaining the process of how it gets from being asked a question to answering it.

“It’s a regulatory issue,” one guest said.

The mortgage market needs joined-up guard rails to AI, one attendee said, saying that firms were currently working on this in their own silos.

“Use AI how you can, don’t use it if you don’t have to and be careful of your data,” a guest warned.

“There is AI FOMO (fear of missing out) at the moment. But if everybody uses it, then everybody is the same, and has the same efficiency and expense ratio. How do you differentiate? I think you go back to people.”