AI-Generated fake IDs are here. Can your property catch them?

Note: Since this post went live, our data shows that it’s been discovered by individuals searching for terms like “AI generated fake ID” and “realistic fake ID.” That’s not who we wrote it for, but this unexpected traffic highlights the urgency of the issue and reinforces the importance of taking proactive measures to protect your community. 

If you work in multifamily real estate, it’s likely your day-to-day technology tools are increasingly infused with AI. Leasing chatbots, predictive AI lead scoring, preventative maintenance, document extraction—these solutions make us more productive, improve customer experiences, and drive down expenses.  

Alternatively, what happens if bad actors get their hands on this technology?   

The reality is that artificial intelligence is now also a tool used by fraudsters, enabling renters to secure housing under false pretenses. What can property management companies do to prevent it?  

Likely your first chance to catch a fraudster is by verifying their ID. You may scan an ID before a tour, whether onsite or virtual, or embed a tool like CheckpointID into your application or screening process.  

Wherever you’re scanning IDs, you want confidence that the technology you invest in is catching fraud, no matter how sophisticated the fraudster is or how legitimate the ID looks.  

In this article, we’ll discuss how fraudsters are leveraging AI to create realistic fake IDs, then show how effective (or ineffective) different ID verification methods are in catching these fakes.   

What is an AI-generated ID? 

Unlike traditional counterfeit IDs, AI-generated IDs are image files that replicate the look of legitimate government-issued IDs. Because these are digital images instead of physical documents, they are easier to distribute and less detectable through traditional verification methods. These fake IDs are frequently sold on the dark web for as little as $5, making them highly accessible to criminals. 

AI tools can generate both front and back images of IDs, including a photo. Additionally, these fake IDs often have realistic-looking backgrounds to make it look like the ID is sitting on a table, countertop, or blanket. 

Can you catch AI-generated IDs? 

AI-generated fake IDs are primarily used for synthetic identity fraud. They are leveraged by fraudsters, usually online, to help themselves or others secure housing under false pretenses.  

To see how the various ID verification technologies fare against these AI-generated fakes, our partner, IDscan, purchased 200 IDs and verified them through different methods, including: 

1. Optical character recognition (OCR)
2. Image and symbol analysis with front & back checks
3. DMV Check 
4. National Lookup (third-party database checks, other than the DMV) 

The findings? AI-generated fake IDs can be caught, but with varying degrees of effectiveness.

Technique #1: Optical Character Recognition (OCR)

The first technique many tools on the market use for validating IDs rely on a type of AI technology called optical character recognition (OCR). OCR enables the tool to “read” text from an image and convert it into machine-readable text. It can identify inconsistencies in text alignment, size, or fonts compared against correctly formatting (per DMV state standards). It also is able to identify inconsistencies in image lighting, shadow placement, or pixelation that would otherwise go unnoticed by humans. 

OCR, despite being the gold standard in ID verification technology just a few years ago, only detected 24.2% of the AI-generated fake IDs. This was a shocking result of the test, but showcases the lightning fast advancement of AI in recent years, and the importance of evolving our tools for catching fraud. 

Technique #2: Image and Symbol Analysis with Front & Back Checks

The next technique leverages OCR as well, but takes it a step further. It ensures the ID passes state-specific security features, such as multi-colored backgrounds, ghost image, and holographics. crucially, it also verifies that the data on the front of ID matches what is embedded into the barcode on the back. This is important, as state-specific security information within the barcode, making it more difficulty to forge.  

The result? This ID verification method successfully detected 84% of AI-generated fake IDs. While significantly improved compare to basic OCR, these final two techniques, which are not available from most providers, can make all the difference in detecting a fraudulent ID.

Technique #3: DMV Checks 

DMV & National Lookup is a unique ID verification enhancement available in CheckpointID that validates the information on the ID against third-party databases. In at least 40 states, the license is cross-referenced directly with DMV records (for the remaining 10 states, read about the National Lookup method below). 

DMV checks are exactly how they sound. In seconds, the license is cross-referenced against Department of Motor Vehicles (DMV) databases to confirm whether an ID was issued by that jurisdiction, using information like the ID number, issue state, issue date, and listed personal information. 

DMV checks caught 100% of these AI-generated IDs.  

How is that possible? Regardless of how realistic the IDs looked, none of them were issued by the DMV, and therefore were identified as fraudulent. 

Technique 4: National Lookup 

As mentioned above, ten states do not currently allow DMV access to non-government entities. How does the DMV & National Lookup feature work in those cases? 

To cross-reference with trusted sources, CheckpointID executes a National Lookup, which queries public, private, and proprietary data sources. For example, we may validate the address against USPS records, use telecom or utility records to deepen identity proofing, confirm identity with the Social Security Administration, and check against known fraud lists.  

Though less effective than DMV checks, it wasn’t by much. National Lookup had a 99.6% catch rate for the 200 AI-generated IDs. 

AI-Generated IDs Caught by Verification Method

Differences in regional effectiveness 

The quality of AI-generated fake IDs varies by state. According to the IDScan.net 2024 Annual Fake ID Report, New York, Texas, and Arizona are common targets due to simpler security features. Regardless, CheckpointID’s detection software flagged 100% of fake IDs from these states in this test. 

Robust digital ID verification to catch AI-generated fake IDs 

For property manager companies looking to reduce fraud through virtual ID verifications, it’s important to leverage technology capable of catching sophisticated AI-generated fakes. Leading solutions must incorporate advanced algorithms, real-time data analysis, and multi-step verification processes to ensure that identities are legitimate. 

Document verification solutions should identify inconsistencies in images and text, while also analyzing data patterns that may indicate tampering or forgery. Lastly, and most importantly, the test above shows that incorporating third-party checks, such as DMV and National Lookup, is the most successful technique for identifying fake IDs. 

Learn more about CheckpointID and how you can defend your property, people, and portfolio against fraudsters.