AI and Lenders: Who’s Liable if LLMs Err?
Private credit firms are racing to automate credit scoring and underwriting, but recent case law proves lenders remain fully on the hook when models make mistakes. The post AI and Lenders: Who’s Liable if LLMs Err? appeared first on Global Finance Magazine .
As lenders increasingly rely on artificial intelligence for tasks such as scoring borrowers and automating workflows, a crucial question arises: when an algorithm makes a mistake, who is held accountable? Credit risk expert Naeem Siddiqi asserts that it's not the AI itself, but rather the lender that bears responsibility. According to Siddiqi, "if the large language model, or LLM, miscalculates a number or uses a prohibited category like race or religion, then the lender is liable."
This sentiment is supported by a court case in which Air Canada was found liable for an error made by its chatbot, despite arguing that the AI was a separate entity. The ruling highlights the precedent that companies cannot escape liability by outsourcing decision-making to AI.
The rise of AI-related litigation extends beyond chatbots, with copyright suits and discrimination complaints targeting AI developers, lending companies, and software firms. While the technology can perform tasks efficiently, Siddiqi emphasizes that its liability sits with the lender. Regulatory bodies such as the Consumer Financial Protection Bureau and the U.S. Department of Housing and Urban Development have made it clear that lenders must ensure compliance with laws such as the Equal Credit Opportunity Act.
This means that even if an AI algorithm introduced bias or failed to provide legally compliant adverse action notices, regulators will pursue the lender, not the AI company.
To navigate this emerging discipline of AI in lending, technology leaders suggest treating AI as a new employee that requires guidance and thorough review. Rather than replacing human decision-makers, AI should be utilized as a support tool that demands oversight. Lenders must provide audit trails and conduct regular back-testing to demonstrate that AI models do not produce discriminatory outcomes.
Although there is enthusiasm for AI's capabilities, it is crucial to acknowledge its potential for mistakes. Ultimately, the balance between regulatory oversight and efficient workflow will be a challenge for human teams, with the hope that AI will eventually handle the review of other AI's work, allowing humans to focus on more critical tasks.
Written by urgent.news from Global Finance's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
