TL;DR: A fake lease agreement is either a genuine tenancy contract edited after it was signed, or a document assembled to describe a tenancy that never existed. It turns up as proof of address in KYC, proof of rental income in mortgage and DSCR underwriting, and proof of business premises in KYB. In the National Multifamily Housing Council's 2024 fraud survey, 93.3% of apartment operators reported fraud in the prior twelve months and 84.3% of those had seen falsified income documents. Visual review fails because a lease has no issuer: any layout is a real layout, so only the file's history and the records outside it can be checked.
A Lease Proves Whatever the Reviewer Needs It To
A lease agreement is a private contract between two parties, and that is what makes it useful to a fraudster. A bank statement has a bank behind it; a lease has a landlord who may be the applicant's cousin, an LLC formed last month, or nobody at all. The same PDF does three jobs. In KYC it is the proof of address accepted when a utility bill is unavailable. In mortgage and DSCR underwriting it is the rental income the loan is sized against. In KYB it shows a company operates from real premises rather than a mail drop.
The National Multifamily Housing Council's Pulse Survey of apartment owners and managers found that 93.3% had experienced fraud in the prior twelve months and 84.3% of those had seen falsified income documents. Cotality's Q4 2025 Mortgage Application Fraud Risk Index put fraud indicators on 1 in 43 investment-property applications and 1 in 27 multifamily applications, against 1 in 118 overall, and attributed part of the growth to DSCR loans, where the lease is the underwriting.
Three production paths cover most fakes. Template farms sell editable lease forms by state or country; the fraudster adds a real address and an invented landlord, which is why template-built documents share clause text across unrelated applicants. Edited-after-creation starts with a real lease, often the applicant's own, and changes the rent, the term, the tenant name, or the address; everything except the changed field is authentic. AI-generated leases are written from a prompt, signature images are added, and the result has never been near a landlord.
Tells That Still Work, and the Ones That Don't
Fonts, layout, and a notary block tell a reviewer nothing, because leases have no standard format and every template is public. The tells that survive connect the document to something it cannot control. The landlord is the first. If the lessor is an entity, it should exist in the Secretary of State or Companies House register, be in good standing, and have a formation date earlier than the lease date. If the lessor is an individual, that person should own the property or manage it for whoever does. A landlord who exists nowhere but on the lease is the most common tell in template-built fakes.
Dates, signatures, and rent come next. Fannie Mae's rental income verification guidance describes a fabricated lease whose termination date fell before its start date. Subtler versions include a lease executed after the loan application and the same signature image on the tenant and landlord lines. Fannie Mae also lists rent that looks inflated against market as a red flag; the check is a listing search for the same bedroom count in the same zip code. When the rent has been edited, the security deposit and the prorated first month usually still describe the original number.
Statutory content is the tell most reviewers skip. In the United States, the EPA's Lead-Based Paint Disclosure Rule requires leases on most housing built before 1978 to include a Lead Warning Statement and a lessor disclosure. In England, the Assured Shorthold Tenancy Notices and Prescribed Requirements Regulations 2015 require landlords to give tenants the How to Rent checklist, and deposit protection rules require prescribed information, including the scheme's name, within 30 days of receiving a deposit. A lease for a 1962 building with no lead disclosure, or an English tenancy that takes a deposit and names no scheme, was written by someone who never had to comply.
None of these tells catches the file that costs institutions money: the edited real lease. Every clause is genuine, the landlord and property exist, and one number changed. A reviewer reading the page cannot see which one.
How Detection Actually Works
Detection that holds up reads the file's story before it reads the contract. Six classes of signal cover edited-after-creation, generated-from-scratch, and recycled fakes; they are questions to answer about any lease behind a decision, not a recipe for defeating a check.
Production method asks how the bytes were made. A lease exported from an e-signature platform carries that platform's fingerprint; one that last left a desktop PDF editor or a generative pipeline carries a different one. Edited-after-creation is its own finding, separate from generated-from-scratch, because the case note should say which fields moved. Timestamp trail asks whether creation, modification, and submission dates cohere with the execution date. A lease dated in January, generated in August, and uploaded the day the underwriter asked for rental income needs a benign explanation; re-saves create noise, so the trail is evidence to weigh rather than an auto-reject.
Issuer matching asks whether the claimed producer matches the landlord named: a national property manager does not issue leases from consumer software. Consistency is the underwriting check, rent to deposit, term to dates, tenant name to the ID and the bank statement, address to the property record. Model artifacts asks whether generative tooling left traces a viewer does not show, including the structural absences of image-only PDFs. Recycled patterns asks whether this template, landlord name, or signature image has appeared in another applicant's file. One reviewer looking at one lease will not see the fifth copy, which is why loan document fraud detection works across the portfolio.
Cross-Checks Outside the Document
A lease that passes file forensics still has to match the world. The county recorder or assessor shows who owns the parcel; the owner should be the lessor or should have appointed the lessor as manager. The Secretary of State record shows when the landlord entity was formed, whether it is active, and who its registered agent is. In KYB, the premises should match the registered address and should not resolve to a virtual office when the applicant claims a warehouse.
Money is the strongest cross-check. Fannie Mae's guidance asks whether the deposit and rent payments appear on the borrower's bank statements with amounts and dates that match the lease, and flags first months paid by money order or deposited on the same day. A departure property that listing sites show for sale rather than for rent contradicts the lease however clean the PDF is. In KYC, the lease should agree with the identity document and the onboarding record already on file; an address that appears only on the lease is not corroborated by it.
What to Do With a Suspect Lease
A flagged lease is not yet a declined applicant. Separate the findings. A landlord that cannot be found, impossible dates, and missing statutory disclosures are enough to hold the file and request the native export or the e-signature certificate of completion.
The break-test is to reach the landlord through contact details sourced independently of the applicant, meaning the property record owner or the registered agent, and confirm the tenancy, the rent, and the start date. A confirmation from the number printed on the lease is the applicant confirming themselves. Escalate when the document and the records disagree and write the inconsistency the way a SAR narrative needs it: which field, which independent record, what the applicant said when asked. Reject the document rather than the customer when no authentic replacement arrives within policy time. The same discipline applies to a proof of funds letter, which also has no institution behind it.
Check a document in the Watchdoc playground with a native signed lease, clean or edited, and compare the x-ray to the reviewer's notes. The first file is free, no email required.
Where Sphinx Fits
Sphinx scores the lease the way a forensic reviewer would, then shows the evidence. Watchdoc runs six checks, production method, timestamp trail, issuer matching, consistency, model artifacts, and recycled patterns, and returns a verdict with the manipulated region highlighted. Published figures: 94.3% correct verdict, 2.8x more forgeries caught, clean files cleared in under 28 seconds, 1 million documents processed, $0.45 per document with no seats or platform fee. The Watchdoc playground runs the same x-ray free, with a 50-document tier. It is built for edited-after-creation as well as generated fakes, and it does not replace the property record or the phone call.
Frequently Asked Questions
What is a fake lease agreement?
A fake lease agreement is a tenancy contract that misrepresents who rents a property, from whom, for how much, or whether the tenancy exists at all. It can be a real lease edited after signing, a template with an invented landlord, or an AI-generated document, used as false proof of address, rental income, or business premises.
Can a lease agreement be used as proof of address for KYC?
Many institutions accept a signed lease as proof of address when a utility bill or bank statement is unavailable, but it is the weakest document in that category because no institution issues it. A lease should be corroborated against the identity document, the property ownership record, and a second address source before it is relied on.
How do mortgage lenders verify a lease for rental income?
Lenders compare the lease to market rents for the same bedroom count and zip code, check that dates are logical against the closing timeline, confirm that deposit and rent payments appear on bank statements with matching amounts, and contact the tenant or landlord through independently sourced details. Fannie Mae's rental income guidance lists these as standard due diligence.
What disclosures should a real lease include?
It depends on the jurisdiction. In the United States, leases on most housing built before 1978 must include a federal Lead Warning Statement and lead-based paint disclosure. In England, landlords must provide the How to Rent checklist and prescribed deposit protection information within 30 days of taking a deposit. A lease missing the disclosures its jurisdiction mandates was likely built from a generic template.
Can AI generate a convincing fake lease agreement?
Yes. A language model can write a jurisdiction-specific lease with correct clause structure and plausible terms, and signature images are easy to add. Detection depends on signals the page does not show: how the file was produced, whether its timestamps cohere, whether the landlord exists in public records, and whether rent deposits appear in the applicant's bank statements.
.png)
