California’s community colleges are a hotbed of financial aid fraud. In fact, according to the state Chancellor’s Office, about one in four applications for financial aid are bogus: in the past year alone, more than 460,000 requests to the state’s online application system are fraudulent requests from bad actors.
Just what can be done about a system that is polluted to this degree? It is an untenable situation with a very straight-forward answer. And that answer is to let technology do the heavy lifting. If there was ever a more important and immediate use for AI in Education, I do not know what that could be.
Financial aid fraud has been going strong for some time, but really took off during the pandemic, when virtually all classes moved online. It simply became easier to get away with identity fraud when fraudsters no longer had to take a seat in class. However, rather than subside after the pandemic ended, the problem has increased.
State and federal governments pour billions of dollars each year into financial aid for higher education. The Free Application for Federal Student Aid (FAFSA) provides a way for students to access financial aid, and over 85 percent of students in the US do just that. Fraud occurs when individuals fill out the FAFSA form with false information so they can illegally obtain money through the U.S. Department of Education.
The most common types of fraud are:
- Underreporting Income
- Fabricating Qualifications
- Hiding Assets
- Pretending to be Someone Else
- Guardianship Transfers
When fraudsters sense an easy target, losses climb. Three years ago, it was suspected that one in five applications were fraudulent. Now that figure is one in four. That translates to millions of dollars out of the pockets of colleges – millions that would be much better spent on tuition, or books, or salaries, or building maintenance, or athletic equipment, or science labs, or just about anything except a constant stream of riches flowing into the hands of criminals.
Part of the rise lies in the fact that the colleges are getting better at detecting it. Some of the signs are obvious, like classes that traditionally never fill up and suddenly do. Others are more subtle.
In the last two years, over $125 million has been allocated to stop the bleeding. The Legislative analyst’s office reports that the 2022-23 budget provides a total of $100 million ($25 million ongoing and $75 million one-time) for various technology and information security purposes. The ongoing funds are primarily for college cybersecurity staffing, whereas the $75 million one-time is primarily for security network upgrades, general security software, and anti-fraud technology. The main goal of these funds is to enhance information security to protect against enrollment scams and hacking.
One of the reasons experts believe that this figure is escalating is because thieves are using AI to figure things out. However, what’s good for the goose is also good for the gander. Catching these thieves is a task that is custom made for AI. Why? Because it is all about data – putting large amounts of data together from disparate sources to catch the bad guys. And nothing excels at this like LLM AI systems.
One of the players getting into the ant-fraud game in a big way is N2N systems, whose Founder and CEO Kiran Kodithala has been leading the AI Foray into higher education with a product called LightLeap, an AI-driven platform that transforms higher education by integrating various data systems to improve learning outcomes, streamline administrative tasks, and enhance decision-making processes.
At its core, LightLeap employs machine learning and natural language processing to interpret and act on vast amounts of data. It integrates with existing Student Information Systems (SIS) and Learning Management Systems (LMS) to pull relevant information, which it then analyzes to identify patterns and predict outcomes. The platform adapts and improves its predictive capabilities over time through deep learning techniques, ensuring tailored interventions and streamlined processes.
According to Kodithala, “Dealing with fraudulent applications using stolen PII Information (SSNs, Driver’s License numbers, etc.) in the California Community College System is a significant challenge.
“This can be a problem for any organization, particularly in educational institutions, where such fraud can lead to substantial financial losses and administrative issues. N2N’s Lightleap.ai is built on proven strategies and best practices that are currently helping Foothill-DeAnza Community College District and West Valley-Mission Community College District detect and mark potentially fraudulent applications as they are submitted.”
This is how it works:
LightLeap Analytics Modules: LightLeap.ai’s machine learning algorithm and analytics modules analyze application patterns and flag suspicious activity. These could include unusual application volumes from similar sources, vector analysis, mismatched geographical data, and patterns that match known fraudulent activities.
Database Cross-Checking: In this step, N2N will cross-reference application data with the customer’s SIS and CRM databases to clear students in good standing and halt students with any preexisting holds. Next, they run the applicant profile against N2N’s proprietary blacklist and whitelist databases to check for inconsistencies or previous records of fraud. This might include past application data, financial aid records, or external fraud databases.
Manual Review Triggers: They establish specific criteria that trigger manual reviews. This can include incomplete or inconsistent information, applications from high-risk locations, or the use of flagged IP addresses (e.g., proxies, VPNs).
Education and Data Model Training: Regularly train staff involved in the admissions and financial aid processes to recognize signs of fraudulent activity. Ensure that the data models and users are trained periodically as they get more familiar with the latest tactics fraudsters use and understand how to handle suspicious applications.
Collaboration and Sharing: Using lightleap.ai and the N2N Illuminate platform can encourage collaboration between departments within the university and with other CCC schools. Sharing information about fraud patterns, bad actors, and blacklisted primary keys (IP Addresses, SSNs, Phone Numbers, Driver’s License Numbers, etc.) can help in developing more robust detection mechanisms across the California Community College system.
In short, the API will take in Application and Student data, perform a look-up against existing records to see if a student exists and if so, do they have any holds. Students in good standing will be cleared from further modeling. The API will return any holds for students with any record of holds. Any other students’ application will be sent to the AI Model to receive a fraud score. AI API will return the application with the Fraud Category and percentage Fraud Score.
Student Aid applications using bogus identities made around $100 million during the past year. That’s nearly twice what they made during the pandemic. This fraud also zaps available seats and drains funds for deserving learners. For more hardcore criminals, it also gives them an additional backdoor into the institution, where students’ personally identifiable information (PII) is highly marketable. For bad actors, it’s like the gift that keeps on giving.
Because data tends to be siloed, institutions can run themselves ragged trying to put the data together. That’s why an automated system using machine learning and natural language processing, AI to you and me, is the logical solution. Foothill-DeAnza Community College District and West Valley-Mission Community College District are among the districts and individual colleges that are getting serious about finding a solution to the problem. And as fraudsters become more sophisticated, count on players like Kodithala’s LightLeap to become more sophisticated as well. “It’s a learning game,” says Kodithala. “Fortunately, the technology learns and adapts. At the very least, it is a long-term scenario. And I am betting it’s a scenario we will win.”
About the author
Kristina E. Greene is a writer, editor and publisher. She specializes in Education Technology and is currently fascinated with the possibilities of AI and its use in bringing quality education to children and adults in less advantaged areas of the world. Kristina enjoys her three grown sons and her retirement, although she hasn’t quite gotten the hang of it yet.
