Prosecutors Say Fraudsters Used AI Ghost Students to Pocket Federal Student Aid

Somewhere in America, a person with a high school diploma and no college history may be enrolled in a semester of online classes right now. They don't know it. Prosecutors say software may be doing their coursework.

That is the picture the Justice Department painted Thursday when it unveiled the first results of Operation Ghostbusters, a nationwide probe into "ghost students." Prosecutors have charged five people and seized more than $750,000 in cryptocurrency from a network associated with Kenyan nationals, which DOJ calls its first international seizure of this kind.

The alleged playbook is simple. Fraudsters take stolen identities, file federal aid applications in those names, and enroll them at colleges and universities. Once loans and grants are disbursed, DOJ says, the refunds are diverted to accounts controlled by foreign actors, with U.S.-based money mules moving the cash.

The AI angle is where it gets interesting, and slippery. DOJ's release says fraudsters use "super intelligence" to place fake students in classrooms and to complete required coursework. It names no tool, no model and no vendor. At a press conference, Assistant Attorney General Colin McDonald of the National Fraud Enforcement Division went further, according to Ars Technica's account:

"It looks like the ghost student is actually completing the class." Assistant Attorney General Colin McDonald

McDonald also said some schemes create the appearance of a fake student in an online classroom to convince schools a real person is present. Treat that as the government's characterization for now.

The numbers are big, and they are not all the same kind of number

DOJ says investigators identified 830,840 potentially fraudulent student aid IDs tied to nearly $130 million, and recommended that the Education Department disable them. Note the word potentially. The FBI's Heith Janke said the bureau has identified more than $32 million in losses in similar schemes. DOJ's overall estimate runs well over $200 million.

The charges themselves are smaller and more concrete. According to the release:

• A New Jersey complaint says three alleged money mules moved more than $1 million in aid to co-conspirators in Kenya.
• A Virginia grand jury indicted a Kenyan national on Oct. 21 for conspiracy to commit wire fraud and student aid fraud, alleging 700 refund deposits totaling about $1,439,548.
• A Nebraska grand jury indicted a Kenyan citizen on Oct. 19 on five counts of theft of government money and five counts of aggravated identity theft,.

An indictment or complaint is an allegation. DOJ's own release says all defendants are presumed innocent until proven guilty in court.

DOJ says there are now more than 40 open cases across 26 FBI field offices. First Assistant U.S. Attorney Theophani Stamos said, "we are just getting started."

The victims are the part that should bother you most. DOJ says many known victims report having only high school diplomas and are shocked to learn of their attendance at an institution hundreds of miles away. Ars notes that victims often learn years later, when the government comes to collect.

There is a catch on the other side of the ledger. The Education Department now relies on online identity checks, where applicants hold a government ID up to a camera and automated systems assign a risk rating. Anyone rated high risk is rejected. Advocates worry that real students will get caught in the net.

"a lot of low-economic kids may not have secured, for example, a driver's license" MorraLee Keller, National College Attainment Network

Keller's worry is that those students could be delayed while they scramble for alternative ID. FTC Chairman Andrew Ferguson, vice chair of the White House Task Force to Eliminate Fraud, said the approach works: "The computers help us find it, and we keep the money in the agency." The department has not said how many real applicants have been wrongly flagged.

The government is using automated systems to catch what it says is AI-assisted fraud. The people most likely to pay for a false positive are the ones the aid was built for.