Artificial intelligence (AI) is poised to transform the mortgage industry. Although AI models could provide more efficiency for housing stakeholders, the data used to train AI models has the potential to perpetuate racially disparate outcomes. In this report, we review the legal and regulatory barriers to adopting AI models that promote efficient and equal outcomes. Through interviews with nearly four dozen housing industry stakeholders in the federal government, financial technology (fintech) companies, mortgage lenders, consumer advocates, and research organizations, we analyze the current strategies for AI adoption and the efforts to mitigate racial disparities.

Our report has two key findings. First, stakeholders have already adopted AI into their marketing, underwriting, property valuations, and fraud detection. Second, smaller, mission-oriented lenders—such as minority depository institutions and community development financial institutions—have lower adoption rates than larger mortgage lenders and government-sponsored enterprises. This technology gap has the potential to reinforce racial disparities in the mortgage market. Based on these results, we provide recommendations to housing stakeholders to (1) design AI models with intention, (2) use pilot programs to measure consumer outcomes under AI models, and (3) produce clear federal guidelines for AI adoption and its consumer protections.