Disclaimer: This content is for educational and entertainment purposes only and does not constitute legal advice. AI-generated content should always be independently fact-checked. You are solely responsible for ensuring your compliance with all applicable laws and regulations. For legal guidance specific to your situation, consult a bar-certified attorney licensed in your state.
What Deepfakes Are and Why Real Estate Is a Target
A deepfake is synthetic media generated or manipulated by AI to convincingly depict something that did not happen. An AI-generated voice that sounds exactly like your client. A video call where the person on screen looks and moves like a real human but is entirely fabricated. A government ID with a face that matches the fabricated video.
Real estate is a prime target for deepfake fraud for three reasons:
- High-value wire transfers: The average home transaction moves $300,000+ in a single wire. That is a life-changing amount of money concentrated in a single moment.
- Multiple handoffs: A typical transaction involves agents, attorneys, title companies, lenders, and clients. Each handoff point is a vulnerability. Each communication channel is an attack vector.
- Time pressure: Real estate runs on deadlines. "We need to wire by 3 PM or we lose the house" creates urgency that overrides caution. Fraudsters exploit this.
The FBI's Internet Crime Complaint Center (IC3) has reported real estate wire fraud losses exceeding $446 million annually, and the trend continues upward—driven largely by AI-enhanced fraud techniques that make social engineering attacks more convincing.
The Three Deepfake Threats in Real Estate
1. Wire Fraud via Spoofed Video Calls
This is the scenario that keeps title companies up at night. Here is how it works:
A criminal compromises the email account of an attorney, title agent, or seller. They study the communication patterns—how the person writes, when they send emails, their typical sign-off. Then, at a critical moment in the transaction, they send modified wire instructions.
When the buyer's agent calls to verify, the criminal uses an AI-generated voice clone or a deepfake video call to impersonate the compromised party. The voice sounds right. The face looks right. The "verification" confirms the fraudulent instructions.
The money wires to the criminal's account. By the time anyone notices, it has been moved through multiple accounts and is unrecoverable.
Cases are emerging where title companies lose six- and seven-figure sums after agents verify wire instructions via video calls with what appears to be a trusted party—but is actually a real-time deepfake. The technology to generate these convincing video impersonations is now available for under $50 per month.
2. Synthetic Identity Fraud
AI can generate entire identities from scratch: realistic faces that belong to no real person, matching government IDs, social media profiles with years of activity, even credit histories built using synthetic identity techniques.
In real estate, this shows up as:
- Fake buyers in cash transactions: A "buyer" with AI-generated identification purchases a property, then disappears after exploiting the transaction in some way (usually as part of a money laundering scheme)
- Fraudulent sellers: Someone impersonates a property owner—typically of a vacant or investment property—using AI-generated IDs and deepfake video to "verify" their identity, then sells a property they do not own
- Mortgage fraud: AI-generated identities with fabricated financial documents to secure mortgages on properties that serve as fronts for financial crimes
3. Manipulated Listing Photos
This is the gray area. There is a spectrum from legitimate virtual staging to outright deception, and AI has made the deceptive end of that spectrum trivially easy.
Legitimate: Virtually staging an empty room with attractive furniture to help buyers visualize the space. The room's dimensions, condition, and features are accurately represented.
Deceptive: Using AI to remove visible water stains from a ceiling, replace a cracked foundation wall with a smooth one, add landscaping that does not exist, or make a 1970s kitchen look like a 2024 renovation. These alterations misrepresent the property's actual condition.
The line is this: does the AI alteration change a buyer's understanding of what they are actually getting? If yes, it is deception regardless of whether you call it "enhancement."
California's AB 723 draws this line legally. Other states will follow. But the ethical obligation exists everywhere, right now, regardless of legislation.
The OODA Loop as a Deepfake Detection Framework
We teach the OODA Loop as a verification framework for all AI output. It is especially powerful for deepfake detection because it creates a systematic process that overrides the emotional urgency fraudsters rely on.
Observe: Notice the Anomalies
Train yourself to notice what is slightly off. In deepfake video calls:
- Unnatural eye blinking patterns (too regular or too infrequent)
- Lip sync that is slightly delayed or imprecise
- Lighting on the face that does not match the background
- Unusual skin texture or hair edges
- The person avoids turning their head to a full profile
In deepfake audio: unnatural pauses, slight robotic quality in sustained vowels, and responses that feel scripted rather than spontaneous.
In altered listing photos: inconsistent shadows, warped edges near edited areas, repeated textures (AI "paints" over areas using patterns that can look duplicated), and perspectives that do not quite make geometric sense.
Orient: Check Against Known Information
Compare what you are seeing against what you already know. Does this person sound like they did last week? Is this attorney using the same account and phone number from the beginning of the transaction? Do these wire instructions match the title company's standard format?
Fraudsters rely on you not checking. The Orient step forces you to slow down and compare current input against established baseline information.
Decide: Verify or Reject
If anything triggers doubt—even mild doubt—the decision is always to verify through an independent channel. Not "I will check on this later." Right now. Before any money moves or any document is signed.
The cost of a false alarm (a 5-minute callback) is zero. The cost of a missed fraud is catastrophic.
Act: Use Trusted Channels
This is where out-of-band verification becomes critical. Hang up the suspicious call. Open your contacts. Call the person back on a number you verified at the start of the transaction—not a number they just gave you.
If the person is real, they will understand. If they are not, you just stopped a fraud.
California's Disclosure Framework and Deepfakes
California has built the most comprehensive legal framework addressing deepfakes in real estate. Three laws work together:
- AB 723 (effective January 2026): Requires disclosure of AI-altered listing photos and videos. Targets the manipulated listing photo problem directly.
- California AI Transparency Act (SB 942): Broader law requiring AI developers to provide detection tools and disclosure mechanisms for synthetic media used in commercial contexts.
- California Digital Replica Act: Protects individuals from unauthorized AI reproductions of their likeness or voice. Relevant when agents use AI to create video content featuring simulated people, or when criminals create deepfakes of real agents.
Together, these laws mean that any AI-generated or materially altered visual media in California real estate marketing must be disclosed, and any unauthorized use of a person's AI-generated likeness is illegal. Other states—notably Colorado, New York, and Illinois—are developing similar frameworks.
Practical Protection Checklist
Implement these ten practices and you will be ahead of 95% of agents in fraud prevention.
10-Point Deepfake Protection Checklist
- Establish identity early: Meet every party to the transaction in person or via verified video call at the beginning of the relationship—not during a crisis moment.
- Create a verification code word: Agree on a unique code word with your client, attorney, and title company at the start of the transaction. Use it to verify identity on any call involving financial instructions.
- Never trust wire instructions sent via email: Always verify wire instructions by calling a number you have on file—not one provided in the email containing the instructions.
- Use callback verification for all financial actions: If you receive a call with wire instructions or changes, hang up and call back on a known number. Every time. No exceptions.
- Question urgency: "We have to wire right now or lose the deal" is the number one fraud script. Legitimate parties understand verification delays.
- Require multi-party confirmation: For any wire over $50,000, require confirmation from at least two independent parties (e.g., title company AND attorney) via separate phone calls.
- Verify identity documents in person: For sellers you have not met face-to-face, require notarized identity verification with an in-person notary who checks physical ID.
- Monitor listing photos for manipulation: If you receive listing photos that look unusually perfect, ask for unedited originals. Reverse image search suspicious photos using Google or TinEye.
- Use encrypted communication: Move sensitive transaction discussions off standard email to encrypted channels (Signal, encrypted email) to reduce interception risk.
- Document everything: Log verification calls, note the time, who you spoke with, and what was confirmed. This audit trail protects you if a dispute arises.
When Enhancement Becomes Deception
As agents who use AI daily, we need to be honest with ourselves about the line between helpful enhancement and misleading manipulation.
These are generally acceptable:
- Virtual staging of empty rooms (with disclosure)
- Brightness and color correction of dark photos
- Removing personal items (family photos, personal effects) for privacy
- Blue sky replacement on an overcast shoot day (with disclosure)
These cross the line:
- Removing visible damage (cracks, stains, mold)
- Adding features that do not exist (pool, landscaping, upgraded finishes)
- Making rooms appear larger than they are
- Changing the view from windows
- Removing neighboring structures or overhead power lines
The test is simple: will a buyer be surprised when they walk through the door? If the answer is yes, you have crossed the line from staging to misrepresentation.
Frequently Asked Questions
How are deepfakes being used in real estate fraud?
The three primary vectors are wire fraud via spoofed video calls (criminals impersonate attorneys or title agents to redirect closing funds), synthetic identity fraud (AI-generated IDs and deepfake video used to impersonate property owners in fraudulent sales), and manipulated listing photos (AI used to misrepresent a property's condition). Wire fraud is the most financially damaging, with losses exceeding $446 million annually.
Can AI detect deepfake property photos?
Detection tools exist—Hive Moderation, Sensity AI, and Microsoft's Video Authenticator can identify AI-generated content with moderate accuracy. However, detection always lags behind generation technology. The best defense is process-based: require original photos alongside any enhanced images, verify property condition through in-person visits, and disclose all AI alterations as required by law.
What is out-of-band verification?
Out-of-band verification means confirming identity or instructions through a completely separate communication channel. If you receive wire instructions via email, verify by calling a phone number already on file—not one from the email itself. If you receive a video call that seems off, hang up and call back on a known number. This technique breaks the fraud chain because criminals typically control only one communication channel at a time.
Do I need to worry about deepfake clients?
Yes. Synthetic identity fraud in real estate transactions has grown dramatically—Sumsub research documented a tenfold increase in deepfake incidents globally between 2022 and 2024. AI can generate realistic faces, matching IDs, and convincing social media profiles. Any transaction involving a party you have not met in person carries elevated risk. This is especially true for cash transactions, vacant property sales, and out-of-state sellers. Require in-person or notarized identity verification for high-value transactions.
How do California's disclosure laws address deepfakes?
California uses three laws: AB 723 (requires disclosure of AI-altered listing media), SB 942 (requires AI platforms to provide synthetic media detection tools), and the California Digital Replica Act (prohibits unauthorized AI reproduction of someone's likeness). Together, they mandate that all AI-generated or materially altered visual media in real estate must be disclosed, and that creating deepfakes of real people without consent is illegal.
Protect Your Business From AI Fraud
Our workshops cover deepfake detection, verification protocols, and compliance frameworks so you can use AI confidently while protecting your clients and your license.