Introducing Fraud Guard 2.0: AI Document Forensics and Behavioral Fraud Detection

When we launched Fraud Guard, the scam playbook looked like this: someone grabs your lockbox code using a fake ID, shows your property to unsuspecting renters, and collects a deposit. Selfie verification, phone number checks, and location gating made that nearly impossible to pull off.

But scammers adapted and rental fraud has gotten more technical. AI image editing tools that once required a graphic designer can now be operated by anyone with a smartphone. The same AI acceleration that has transformed leasing has also lowered the barrier to committing identity fraud. The volume of manipulated IDs showing up in prescreenings is reflecting that.

Fraud Guard 2.0 addresses both the new document threat and a second pattern we have been tracking: behavioral signals that indicate a scammer is trying to gain access to houses in your portfolio, rather than genuinely looking to rent.

What's new in Fraud Guard 2.0

AI-powered forensics on uploaded IDs

The original ID verification layer confirmed that a document looked legitimate and matched the applicant's stated details. What it did not do was examine the image itself for signs of manipulation.

Fraud Guard 2.0 adds a forensic layer that does exactly that. When an applicant uploads an ID during prescreening, RentEngine now performs an image-level analysis looking for:

• Inconsistent lighting or grain around the portrait — a common artifact of face-swapping

• Halo or seam artifacts from digitally pasted portraits

• Moiré patterns that indicate the document was screen-captured rather than photographed directly

• Security feature irregularities that suggest the document has been modified

It also estimates the apparent age of the face in the photo and compares it to the date of birth printed on the document. A mismatch of more than 15 years triggers an automatic rejection.

Rejections are issued when the system detects manipulation at medium or high confidence. Everything is logged to the applicant's record so you have a clear audit trail.

Behavioral fraud flags

Not all fraud involves a fake ID. A second pattern has emerged: prospects whose showing behavior does not match the profile of a genuine applicant.

Properties at dramatically different price points. Multiple showings of the same unit. Showings initiated from hundreds of miles away. Individually, any one of these could have an innocent explanation, but together, they form a recognizable pattern.

Fraud Guard 2.0 now monitors for five behavioral signals that are statistically associated with fraudulent intent:

• Interest in properties whose rents differ by more than 50%

• A second showing request for the same property

• More than 2 prescreening rejections before eventual approval

• A showing initiated from more than 20 miles away from the property

• Interest spread across properties that are geographically inconsistent with your portfolio

Each check and the corresponding thresholds are configurable — you can adjust the price differential percentage, the distance threshold, and the rejection count to match what makes sense for your portfolio.

When any of your triggers fire, the prospect is flagged. All future showing requests require PM approval, and the showing pending event is labeled "Potentially fraudulent activity" so your team has immediate context.

Where to find it: Settings > Showings > Prescreening Template > Fraud Guard > Behavioral Check

What was already in Fraud Guard

Fraud Guard 2.0 builds on top of these measures:

ID verification — RentEngine verifies that uploaded IDs are real and valid documents, checks for biometric discrepancies, and runs them against a global blocklist. You can also decide whether an ID can be uploaded from the prospect’s files/camera rolls or taken in the moment.

Selfie and liveness check — A live selfie is matched to the ID provided. Because the selfie is taken live through the front-facing camera, a printed photo or a screenshot cannot be used. The person booking the showing has to be the person on the ID.

Phone number verification — Fraud Guard validates that the prospect's number is a real, carrier-verified U.S. mobile number. VOIP numbers like Google Voice — a common tool for scammers who want to operate anonymously at scale — are blocked.

Location verification — When a prospect tries to access the lockbox code, their device must be physically at the property. The code cannot be shared remotely. The verification radius is configurable, so you can tighten or loosen the geolocation requirement based on your portfolio's geography.

Source verification — Before receiving an access code, prospects must confirm where they found the listing. Anyone who indicates Facebook Marketplace or Craigslist is prompted to call your team directly, and you are alerted before they ever enter the property. This layer addresses a more recent scam pattern where fraudsters bypass ID checks entirely by sending victims your own scheduling link — the victim passes every verification check, the scammer never appears in your system.

Captive code generation — Lockbox codes are never sent via text message. They are generated through the self-showing flow and are only valid for one hour.

The full Fraud Guard 2.0 stack

Fraud Guard 2.0 now combines seven layers of protection:

1. ID verification and global blocklist

2. Selfie and liveness check

3. Phone number verification (no VOIP)

4. Location verification at showing start

5. Captive, time-limited lockbox codes

6. AI forensic analysis of uploaded ID images

7. Behavioral fraud flagging across your portfolio

Used together, these measures make it significantly harder for a bad actor to get through and give you a documented record when they try.

Fraud Guard 2.0 is live now for all RentEngine customers. Questions? Reach out via the Chat.