Ghosts in the Classroom: How AI Fraud Is Haunting College Campuses

As International Fraud Awareness Week (November 16–22, 2025) marks its 25th anniversary, stories like Marley Stevens’ and the rise of “ghost students” reveal how artificial intelligence is fueling a new wave of academic fraud—one that threatens both innocent students and public trust, says Accounting and Information Assurance Senior Lecturer Samuel Handwerger at the University of Maryland’s Robert H. Smith School of Business.

Stevens thought her freshman year at the University of North Georgia would be defined by exams and late-night study sessions. Instead, her semester unraveled after a professor accused her of cheating—based solely on Turnitin’s AI-detection score. The software flagged her paper as AI-generated, but Marley had only used Grammarly, a school-recommended writing tool.

“While Marley fought to clear her name, a darker form of academic deception was unfolding across America’s community colleges—one that didn’t involve students cheating with AI, but AI itself pretending to be students,” notes Handwerger, who advises UMD’s Justice for Fraud Victims chapter.

Rise of the “Ghost Students”

Organized crime rings are now using AI bots to impersonate students, enroll in online classes and siphon off federal financial aid, Handwerger says. In California alone, over 223,000 fake enrollments and $11.1 million in unrecoverable aid were uncovered in 2024. These “ghost students” use stolen or fabricated identities to apply for college, get accepted, and vanish once the aid hits their accounts. “They even ‘attend’ class—submitting AI-written homework—to appear real long enough to cash out.”

A significant factor is the “sockpuppet”—a fake online persona created by AI apps. A typical sockpuppet account, Handwerger says, often looks complete because it uses assets from readily available places—for example, a social-media profile like Twitter/X or Instagram, a disposable or free email address, a generic online résumé or profile on a directory site, and an image sourced from stock-photo libraries or AI avatar generators.

But Handwerger says those “surface signals” don’t prove identity. “Educators and investigators should treat such combinations as red flags rather than proof,” he advises. “Look instead for inconsistencies, verify with primary documents or institutional records, and report suspected identity theft to authorities and resources like the FTC or your institution’s security/financial-aid office.”

At one college, 50 fake applications were submitted within two seconds. In another case, a professor thought her class had finally filled to capacity, only to learn that nearly all her “students” were algorithms. “The scope is staggering,” Handwerger adds. The U.S. Department of Education estimates $90 million in aid has been stolen, some of it even using the identities of deceased individuals.^”

Two Sides of the Same AI Coin

The two stories—Marley’s false accusation and the ghost student epidemic—illustrate how AI can both wrongfully convict the innocent and empower the guilty, Handwerger says. “Forensic accountants, fraud investigators and higher education administrators now face a dual challenge: protecting honest students from false AI-detection flags and identifying fraudsters who weaponize AI to commit large-scale financial crimes against colleges and taxpayers.”

What Forensic Accounting Can Teach Us

From a forensic perspective, these schemes leave behind digital fingerprints—patterns of addresses, email domains, and submission timing that betray nonhuman coordination. Data analytics tools, Handwerger says, can reveal and verify clusters of fraudulent activity by comparing multiple FAFSA applications from the same IP address or phone number, course submissions with identical AI-generated language patterns, and login timestamps that occur at impossible intervals. “This is modern forensic work: not chasing paper trails, but algorithmic trails.”

The Human Cost Behind the Data

Victims like Heather Brady of San Francisco discovered $9,000 in loans taken out in her name for classes she never attended,” Handwerger says, adding: “Brittnee Nelson, a Louisiana small-business owner, spent two years untangling fake student loans that nearly went to collections. Meanwhile, students like Marley Stevens lose mental health, scholarships and trust in the academic system—all because of AI systems gone awry.”

For students of fraud and forensic accounting, Handwerger says, these cases exemplify:

• Identity theft and synthetic fraud—when identities (real or fabricated) are used to extract benefits from institutions. Verify the application details.
• Control weaknesses—especially in open-admission or automated systems where identity verification is weak.
• Data analytics for detection—using outlier analysis, Benford’s Law, and duplicate-address detection to reveal clusters of fraud.
• Ethical oversight— the importance of human judgment in AI-assisted decisions.

“Every forensic accountant,” he says, “must now be both a data analyst and, to a certain extent, an ethicist, at least because suspicion should be attached to everything.”

A Call to Action

The U.S. Department of Education is now requiring government-issued ID verification for first-time aid applicants, but the arms race continues. Regarding future fraud examiners, Handwerger advises that students should see this as a case study in real-world digital deception—one that merges psychology, cybersecurity, and forensic finance. “Whether it’s an innocent student accused of cheating, or a fake one stealing financial aid, AI has blurred the line between human and machine fraud,” he says. “And on today’s college campuses, sometimes the most dangerous ghosts aren’t haunting the dorms—they’re haunting the databases.”