Fraud got cheaper. Screening didn’t keep up.

A convincing fake pay stub now costs an applicant a few dollars and a few minutes to generate. The FBI just put a number on what that’s doing at scale: $275 million in reported real estate fraud losses in 2025, up 59% from $173 million the year before, across 12,368 victims. And that’s only what got reported — rental fraud routinely stays off the books because managers absorb the loss or don’t spot it until after an eviction.

Here’s the part that should bother you more than the number: the tools to fake income scaled faster than the tools to verify it. Most screening processes were built for a world where forging a document took real effort. That world is gone, and a leasing agent eyeballing a PDF is no longer a control — it’s a formality.

The same asymmetry runs through everything in this issue. Fraudsters are using AI at full production depth. Most operators our size are still in test mode. That gap is the story of 2026, and it’s widening every quarter you stay in pilot.

Also: HUD says you own your leasing chatbot’s answers, Apartments.com replaces the filter grid with an AI advisor, and a free two-minute paystub check your team can run this week.

The Radar

FBI logs $275M in rental fraud: The FBI’s Internet Crime Complaint Center counted more than $275 million in real estate fraud losses in 2025, up from $173 million the year before — a 59% jump across 12,368 reported victims. The category covers rental and timeshare scams, and the report flags AI-driven schemes like deepfakes and voice cloning as the fastest-growing tools. The real number is higher: rental fraud routinely goes unreported because managers absorb the loss or don’t spot it until after an eviction. What it means: Fake pay stubs now cost applicants a few dollars to generate, so if your screening still leans on documents a leasing agent eyeballs, budget for income verification that pulls from the source — one bad approval costs more than a year of the tooling.

AI leasing tools carry FHA liability: HUD’s guidance on algorithmic tools keeps sharpening its teeth. Disparate outcomes are actionable regardless of intent, and “the vendor built it” is not a defense — SafeRent paid $2.275 million after its screening algorithm scored out Black and Hispanic applicants and voucher holders, and Harbor Group got sued when its chatbot screened out voucher holders on its own. Enforcement has shifted too: private fair housing nonprofits processed 74% of complaints in 2024, and they now test leasing chatbots remotely and at scale. If an AI agent answers your prospect calls, it’s answering to the Fair Housing Act. What it means: Audit your leasing bot’s responses to voucher, disability, and familial-status questions now — you own its answers, not the vendor.

Apartments.com goes conversational: On June 16, CoStar launched Apartments.com Ai, replacing filters and keyword search with a chat-style advisor for every renter on every visit — an upgrade of the AI Smart Search it rolled out sitewide in January. The AI answers property and neighborhood questions, compares communities side by side, and walks renters through 3D tours before they ever contact you. That means an algorithm, not a filter grid, now decides which listings get surfaced. What it means: Thin listings lose — audit your Apartments.com data for complete amenities, current photos, and accurate pricing, because the AI can only recommend what you’ve actually filled in.

The Breakdown

The AI gap is no longer about access. The tools got cheaper and simpler this year, and the gap between large and small operators got wider anyway — which means waiting for AI to become “easier” is a losing strategy.

The measurement, not the forecast. Eurostat data shows the AI adoption gap between large and small EU enterprises grew from 30 percentage points in 2024 to 38 points in 2025. In April, PwC found that just 20% of companies are capturing 74% of all AI-driven economic value, with the top fifth pulling 7.2x more AI revenue than everyone else.

Adoption isn’t the bottleneck — depth is. Small-business AI adoption hit 66% this year, up 11 points, yet 70% of owners say they need more training to use it effectively. Pax8’s Q2 survey of 400+ SMB leaders found the real divide is between businesses deploying AI in production and a “stuck middle” that can’t move beyond testing. Census Bureau data shows the same barbell: enterprises pull away, solo operators surge, and the middle stalls. A 100–2,500 unit portfolio is the middle.

Why it compounds. This issue’s theme is the proof. Fraudsters already use AI at full production depth — a convincing fake paystub costs a few dollars and minutes. The large operators answer with portfolio-wide detection tooling. An operator still “piloting” AI is running last year’s screening against this year’s fraud, and the distance widens every quarter you stay in test mode.

The move: stop sampling tools and pick one workflow — screening document review is the obvious candidate — then deploy it across every property, train the team on it, and measure the result for 90 days. One workflow at full depth beats five pilots. The stuck middle isn’t stuck for lack of tools; it’s stuck for lack of a decision.

This Week’s Move

This week, open the last ten PDF paystubs your team approved and check each file’s document properties (File > Properties in any PDF reader). A real payroll PDF’s “Producer” field names the payroll system — ADP, Paychex, Gusto. Generated fakes usually show a generic PDF editor or an online paystub builder. Takes about two minutes per file, costs nothing, and no vendor is involved. The National Apartment Association’s research makes the uncomfortable point: AI-generated documents are beating AI detection, so process is the layer that still works. Second rule to adopt while you’re in there: verify employment by calling a number you look up yourself — company website, not the one typed on the application. Applicants control everything on the form; they don’t control Google. Run both checks on every applicant, no exceptions — uniform application is what keeps this clean under Fair Housing.

Worth a Look

Bisnow’s “Apply, Lie, Move In.” The clearest read yet on how AI-generated pay stubs and bank statements are beating standard document review — including the Los Angeles landlord who found out after move-in — and it’s the piece to circulate if your team still trusts a PDF on sight: AI Is Making Fraudsters Harder To Spot. Apartment Owners Are Paying The Price.

Multifamily Dive on AI leasing’s legal exposure. Three federal cases, a HUD guidance memo, and a 20% jump in ADA lawsuits have shifted the compliance math on AI leasing agents — if a chatbot answers your leasing line, this is the 10-minute read that tells you what to ask your vendor this week: The hidden legal risk of AI apartment leasing tools.

Commercial Observer’s fraud-detection tooling survey. A useful, vendor-plural map of how landlords and lenders are verifying income and identity now that fake documents are cheap to make — good starting point if you’re evaluating whether your screening stack needs a verification layer, not an endorsement of any one tool: Proptech Rides to the Fight Against AI-Driven Rental and Mortgage Fraud.

The fraud got cheap and the tools got easy — the only thing still expensive is the decision to run one of them at full depth, so make that decision this week.

— Josh