In September 2025, a man calling himself Igor was approved for two apartments in Hollywood, each renting for $5,300 a month. His bank statements, proof of employment and ID all cleared. Two days later, the landlord's bank flagged the cashier's checks for first and last month's rent as fraudulent.
When the owner, Michael Renkow, got to the building, one door was pinned shut from the inside. Meanwhile, in another unit, a woman who had paid Igor to sublet told him she had established residency and wasn't leaving.
The eviction took seven months, and Renkow said the whole episode cost him about $90,000. In May 2026, he was thinking about selling the building altogether.
If a bank had onboarded a customer that way, it would be a compliance scandal. But in rental housing, it barely made the news, and the only reason we know how it ended is that the suspect used his real face on the fake ID. Police eventually tracked him down, but not until January this year--four full months after the initial lease was signed.
Leasing is one of the least protected onboarding checkpoints in the financial system, and AI has found it.
A credit decision nobody deems as one
Think about what a lease actually is. A 12-month lease at $5,300 a month is roughly $63,600 in rent extended on trust, plus the keys to an asset worth far more. That's an unsecured credit decision by any reasonable definition, and yet, it's usually made by a leasing agent with a deadline, looking at PDFs.
The deadline is part of the problem. Chris Rankin, CEO of screening company Rent Butter, told Bisnow that tighter competition and ever-larger portfolios are pushing property managers to fill units faster, making it hard to look closely at every application.
In a lender's world, speed and scrutiny are traded off deliberately, whereas in leasing, speed usually wins by default.
Banks have spent years hardening this exact moment. In November 2024, in fact, the U.S. Financial Crimes Enforcement Network (FinCEN) issued an alert warning financial institutions that criminals were using generative AI to fake identity documents and get past verification controls.
Most financial institutions responded with the tools they already had. Think customer identification programs, liveness checks, suspicious activity reports, and a regulator watching. When one bank flags a fake identity, that signal goes somewhere; a landlord who spots a forged pay stub usually just rejects the applicant, who is free to try the building down the street.
Leasing does have a federal rulebook, but it points in a different direction. Under the Fair Credit Reporting Act, the FTC treats tenant screening companies as consumer reporting agencies, with a duty to keep their reports accurate. This protects applicants from being wrongly denied because of bad data, which is important but doesn't tell a landlord how to catch an applicant whose data is fake.
That part is left to each operator's budget and judgment. Yet, a leasing office has an inbox and a vacancy target.
The fakes got cheap
Most people underestimate how easy this has become. MRI Real Estate Software, for example, bought 200 AI-generated fake IDs, some for as little as $5, and tested them against standard verification tools, finding optical character recognition readers, the industry default, caught only 24%.
Recently, a Checkr survey of 1,400 landlords across 14 U.S. cities concluded that, in San Francisco alone, 85% of them are worried about application fraud. One in three had discovered identity theft in their buildings, and 70% had discovered or suspected application fraud that ended in an eviction.
Most respondents actually found out only after the tenant had moved in. At the worst possible time to learn anything.
Industry-wide, a National Multifamily Housing Council survey found that respondents wrote off an average of nearly $4.2 million in bad debt over 12 months, and about a quarter of it was tied to nonpayment by tenants who got in with fraudulent applications.
Fraud also runs the other way. The FBI's 2025 Internet Crime Report logged 12,368 real estate fraud complaints and $275 million in losses, a category that includes fake listings aimed at renters.
Where losses land
When a fraudulent tenant gets in, the cost doesn't stay with one landlord, but shows up as bad debt, legal fees, damaged units, and months of lost rent. Then, it spreads.
Owners are already squeezed on insurance. Property insurance costs for securitized multifamily properties rose 58% over five years, according to New York-based analytics firm Trepp, making it their fastest-growing major expense.
A tenant who lied their way in is also unlikely to carry real coverage, so when something goes wrong, the damage and liability tend to fall on the owner's policy. That feeds loss history, renewals, and eventually the rent everyone else in the building pays.
Deposit alternatives and liability programs have to price in those losses, too, meaning honest renters end up subsidizing those dishonest ones.
But fintech has already worked out that you can't win an arms race over what a PDF looks like. Lenders moved to verifying at the source, with direct payroll connections, bank data, and continuous signals instead of one-time snapshots.
Leasing should do the same.
The good news is that a lot of source-verified data already moves through the leasing workflow, but it just isn't treated as risk data. Does a renter's insurance policy actually bind at lease signing? Is the "third-party policy" someone uploaded real, and does it stay active? Did the deposit payment clear through a traceable instrument?
A fair objection
Consumer advocates would push back on the story as told so far, and they have a record to point to. Much of the fraud data, including some of the figures above, comes from landlord trade groups and screening vendors, who have a commercial interest in the problem looking big.
They also have evidence that more screening doesn't automatically mean better screening. The National Consumer Law Center's Digital Denials report, for example, found that tenant screening reports often carry inaccurate eviction and criminal records, that landlords frequently rely on a single automated score, and that the harm falls disproportionately on Black and Latino renters.
Regulators have acted on this before. In 2023, the CFPB and FTC asked a federal court to order TransUnion to pay $11 million to consumers plus a $4 million penalty over its tenant screening practices. The CFPB has also noted that there is no proof screening reports actually predict how someone will behave as a tenant.
Then, there's the cost. The FTC is now examining rental fees across the whole lease, from application to move-out, while NCLC separately argued that some proptech products, like security deposit alternatives, end up making renting more expensive. If every new fraud check becomes another fee on applicants, honest renters pay for a problem they didn't create.
That's a fair point, and it shapes what a good answer should look like. Verifying at the source and reusing signals that already exist in the lease lifecycle should make leasing cheaper and fairer for the honest majority, not more expensive.
Banks learned long ago that onboarding is where fraud is cheapest to stop. Rental housing is learning the same lesson one eviction at a time, and landlords far and wide are paying for it.
