AI mistakes leave lenders facing liability questions
Private credit firms are shifting more of their underwriting and portfolio monitoring to artificial intelligence. When the algorithms make mistakes, the lenders—not the AI—will bear the consequences.
Over half of private credit portfolio managers—54%—plan to use AI in underwriting within the next year, based on a March survey of 120 global firms. The transition is already happening, with lenders employing large language models to assess borrowers, track risks, and automate tasks that previously required weeks of analyst work.
Credit risk expert Naeem Siddiqi, author of Intelligent Credit Scoring and a senior advisor at SAS, stated that firms cannot escape responsibility. “If the LLM miscalculates a number or uses a prohibited category like race or religion, the lender is liable,” he said.
The legal system has already addressed this issue. In Moffatt v. Air Canada, a customer used the airline’s chatbot in 2022 to ask about bereavement fares after a family death. The chatbot provided incorrect information, telling him he could book a full-fare ticket and request a refund within 90 days, though the airline’s policy required requests before travel. When the customer sought reimbursement, Air Canada argued the chatbot operated independently.
The British Columbia Civil Resolution Tribunal rejected that argument. The airline was ordered to pay $812.02 Canadian dollars in damages, interest, and fees. The decision established that companies cannot avoid liability by claiming their AI acts alone.
Related: Infantino’s World Cup Overhaul Shows Leadership Bias
Regulators have made it clear that lenders remain responsible for compliance, even when using third-party software. The Consumer Financial Protection Bureau, the Office of the Comptroller of the Currency, and the U.S. Department of Housing and Urban Development have all indicated that if an AI algorithm introduces bias, violates the Equal Credit Opportunity Act, or fails to provide required adverse action notices, the lender—not the AI provider—will face penalties.
Omar Abassi, founder of lending tech startup LoanFlo AI, said the legal responsibility is clear. “The buck stops entirely with the lender,” he explained. Some firms now require vendors to supply audit trails and regular back-testing to ensure models do not produce discriminatory results.
This balance between AI adoption and risk management is influencing how lenders implement the technology. David Yahalomi, COO and co-founder of loan-management platform Hypercore, advised against giving AI final decision-making authority. The technology relies on statistics and is not error-proof. Mistakes will happen, and lenders must prepare for them.
Abassi described the situation as unsustainable. “Underwriters are taking on too much work in traditional setups, which increases the chance of human error,” he said. Still, he does not expect AI to fully replace them soon. In the future, AI may review its own work. “AI agents will eventually check each other’s output,” Abassi predicted.
Firms looking to refine their cash management strategies can explore top providers for 2026 to complement their AI-driven lending processes.
