A vacant land transaction in Maryland nearly became the latest victim of deepfake fraud last week when artificial intelligence (AI) was used to impersonate the property owner during a live video session.<\/p>\n
That roughly $100,000 deepfake fraud<\/a> attempt was stopped by identity verification and transaction security platform Proof<\/strong> during a remote notarization, the company said.<\/p>\n The incident highlights a continued vulnerability in real estate transactions, particularly those involving out-of-state sellers or vacant land where in-person verification is impossible.<\/p>\n \u201cReal estate fraud isn\u2019t always a high-volume issue, but it\u2019s an incredibly high-impact issue,\u201d Proof CEO and co-founder Pat Kinsel<\/a> told HousingWire<\/strong>. \u201cA lot of title companies will say, \u2018It hasn\u2019t happened to me,\u2019 and then when it does happen to them, it\u2019s devastating.<\/p>\n \u201cI know a title agent that had a million-dollar loss, and they ended up having to personally cover this.\u201d<\/p>\n Tech-enabled fraud reached $13.7 billion in 2024, according to the FBI\u2019s<\/strong> Internet Crime Complaint Center. Meanwhile, deepfake-related scams are rising quickly \u2014 jumping 40% year-over-year, per Entrust\u2019s<\/strong> 2026 Identity Fraud Report<\/a>.<\/p>\n In a demonstration, Kurt Ernst, product manager at Proof<\/a>, showed how easily commercially available software can create convincing deepfake videos.<\/p>\n Ernst placed a deepfake face over his own in real time, noting the setup took just 15 minutes using off-the-shelf technology.<\/p>\n \u201cI don\u2019t have a supercomputer sitting in my closet here running the latest in video drivers,\u201d Ernst said. \u201cWe set this up very quickly. It\u2019s using commercial, off-the-shelf software that you can get, or your fraudsters, your friendly neighborhood fraudsters, can get as well.\u201d<\/p>\n A recent study<\/a> by Deloitte<\/strong> estimated fraud losses tied to AI-generated deepfakes could reach billions annually, with financial services and real estate among the most exposed sectors.<\/p>\n Ernst also cited common flaws in deepfake videos, including hand warping when crossing the face, imperfections in facial hair rendering and inconsistencies in mouth movement.<\/p>\n \u201cYou can see how there\u2019s warping there. You can see my face over my fingers,\u201d he said. \u201cThese different types of technologies are terrible with fingers and hands, and they\u2019ll look kind of creepy.\u201d<\/p>\n Proof\u2019s technology flagged the deepfake video as fraudulent within seconds during the demonstration.<\/p>\n The system scans video frames in real time while also analyzing device information, location data and email addresses, Ernst added.<\/p>\n The Maryland<\/a> transaction involved a vacant lot where the seller claimed to be local but, as often is the case, was operating elsewhere.<\/p>\n \u201cThey don\u2019t always have the exact details on their ID correctly,\u201d Ernst said of scammers<\/a>. \u201cTheir email address \u2014 that\u2019s a classic one. You can spin up a new email address in five seconds. So their email address will have never been seen by anyone.\u201d<\/p>\n Detection systems can flag suspicious transactions and red flags autonomously.<\/p>\n \u201cWe can say, \u2018We think this one needs to be reviewed,\u2019 even if the notary<\/a> doesn\u2019t see it,\u201d Ernst said. \u201cThat can happen almost instantaneously.\u201d<\/p>\nHow deepfakes evade human detection<\/h2>\n
Multilayered detection approach<\/h2>\n
Best practices for professionals<\/h2>\n