Artificial intelligence is beginning to play a positive role in the mortgage origination process – from applications to verifications to valuations to closings and beyond, it is helping to speed processes and reduce the rate of error.
But AI can be a double-edged sword, as there is also a rising risk of AI-driven fraud.
In a recent report, Sarah Frano, vice president and real estate fraud expert at First American Title Insurance Company, explains that scammers are increasingly harnessing AI to generate deepfake videos and fraudulently pose as legitimate borrowers.
“Detecting deepfakes can be challenging, but there are several techniques and tools that can help identify them, including visual and behavior analysis,” Frano says in the report. “Look for inconsistencies in blinking patterns, lip movements, reflections, shadows, skin texture, and hair. Deepfakes often struggle to replicate natural blinking patterns and lip movements and may show overly smooth skin or inconsistencies in hair.”
A recent study from Deloitte estimates that generative AI could cause U.S. fraud losses to grow by 32% each year moving forward, reaching $40 billion by 2027.
So how can lenders and borrowers protect themselves from deepfake scams during the home buying and selling process?
“Always verify the identity of the person you are dealing with through multiple sources,” Frano says. “Whenever possible, meet in person to confirm details and verify identities.”
Another key step is to always use trusted platforms.
“Conduct transactions through trusted escrow services and their secure platforms,” she says. “Be cautious of emails and content from unknown or unverified sources.”
Protecting the title, of course, is a key part of the strategy.
“Where available, always purchase a title insurance policy that covers fraud after you purchase your home,” Frano advises.
In addition, lenders and borrowers need to “keep up to date with the latest fraud schemes and how to detect them. Educate yourself and others about the signs of deepfakes and encourage a critical approach to consuming digital content,” she says.
“By combining these techniques and staying vigilant, you can improve your ability to detect deepfakes and reduce the risk of real estate fraud powered by AI,” Frano says.
So how are scammers using AI to commit real estate fraud?
“AI tools also make it easier to quickly fabricate correspondence, identification, deeds, mortgages, video, and voices, which can be indistinguishable from a real document or person,” Frano says. “Given the intrinsic value of real estate, property transactions and mortgages are attractive targets for scammers.”
“Armed with AI, scammers can commit broader and increasingly complex types of fraud,” she says. “Deepfakes, for example, are created using deep learning algorithms that gather a large dataset of images or videos of a target person, which is then used to train an AI model to understand the person’s voice, facial features, expressions, and movements. In real estate transactions, scammers can use deepfake audio or video to impersonate real estate agents or other professionals involved in the transaction, leading to fraudulent communications that provide false information or instructions.“
Scammers can also use deepfakes to impersonate home sellers.
“Recently, a Florida title company scheduled a video call to confirm the identity of a woman attempting to sell a vacant lot,” Frano says. “They were shocked when they encountered an AI-generated person. The fraudsters likely used AI and face-swapping technology to create the alleged seller, but the face was actually that of a woman who had disappeared in 2018.”
The threat of AI-driven fraud will be discussed in detail in an upcoming First American webinar, “Unreal Deals: AI, Deepfakes and How Your Closings are at Risk.”