Get all your news in one place.
100’s of premium titles.
One app.
Start reading
Fortune
Fortune
Luisa Beltran

Frank’s Charlie Javice vs. JPMorgan: Why the case could hinge on “synthetic data"

(Credit: Courtesy of Frank)

It can help detect fraud, gain efficiencies, and even power self-driving cars. And so-called synthetic data also happens to be at the core of a case that has pitted a young fintech entrepreneur, accused of using the tech tool, against JPMorgan Chase, one of the world’s biggest banks, which bought her student loan company, Frank, and later sued her for fraud.

In early April the case against entrepreneur Charlie Javice, 31, was back in the headlines after she was arrested amid criminal charges filed by the Department of Justice. The DOJ alleges that the former Frank CEO committed fraud when she sold Frank to JPMorgan. Javice “falsely and dramatically” inflated the number of customers Frank had in order to induce JPMorgan to buy the company, according to the U.S. Attorney’s Office for the Southern District of New York, which is part of the DOJ. The Securities and Exchange Commission separately sued Javice, claiming she concocted a fraudulent scheme to hide the fact Frank had identifying data for only about 300,000 students, much lower than the 4.25 million customers she repeatedly touted, the April 4 lawsuit said.

But JPMorgan has come under the microscope as well. The Office of the Comptroller of the Currency is scrutinizing JPMorgan Chase's due diligence of recent acquisitions, the Financial Times reported April 7. The OCC, which supervises national banks, has scheduled a specific audit of JPMorgan Chase's dealmaking, after the bank acquired dozens of small companies in 2021 and 2022, the story said. This would include its $175 million buy of Frank in September 2021. The OCC and JPMorgan declined comment.

Now the criminal case and the civil cases will play out alongside each other.

On the civil front, JPMorgan Chase alleged in a December lawsuit that Javice, the founder and former CEO of financial aid site Frank, along with Olivier Amar, Frank’s chief growth officer, hired an unnamed data science professor to make up fake, or synthetic, data that they used to trick the bank into believing the startup had 4.25 million customer accounts. But Javice, in her Feb. 27 answers to the JPMorgan filing, claims it was the bank itself that requested she use synthetic data to represent Frank’s more than 4 million users. Javice, who is represented by Alex Spiro (a partner at law firm Quinn Emanuel Urquhart & Sullivan and Elon Musk’s personal attorney), wants $27.9 million in compensation and a trial by jury. “The damage wrought by JPMC’s falsehoods is likely to be lifelong,” the court filing said. JPMorgan is currently seeking to move Javice’s employment counterclaims to arbitration, while Amar wants to dismiss the complaint against him, alleging he was not party to JPMorgan’s merger agreement with Frank and that the bank doesn’t claim he actually engaged in any fraudulent conduct, according to a March 1 court filing. A Delaware Chancery Court judge also heard oral arguments from both sides on March 8 about whether JPMorgan must pay for Javice’s and Amar’s legal fees, but has yet to return a verdict. 

The case of JPMorgan vs. Javice

Javice, in her answers to the JPMorgan filing, painted a picture of a banking giant that rushed to buy a fintech startup without conducting adequate diligence, hoping Frank would give JPMorgan access to a key demographic: young students. She alleged that JPMorgan Chase knew exactly what it was getting when it bought Frank, which claimed to simplify the financial aid process, in 2021 (you can read Fortune’s full account of how JPMorgan fell for Frank here). Javice alleges that Jamie Dimon, JPMorgan’s billionaire CEO, even took a personal interest in the bank’s purchase of Frank, telling Javice in July of 2021 that he thought JPMorgan should “get the deal done.” Dimon also learned there was a rival bank interested in buying Frank, Javice said in her filing. He wanted to close the bank’s buy of Frank before the start of the school year and before the college application season, Javice claims in her answers to the JPMorgan lawsuit. 

At the heart of the dispute is Frank’s use of synthetic data. JPMorgan claims that the startup used the tool to artificially inflate, or fake, the number of users Frank had. But Javice claims that it was JPMorgan who asked Frank to provide the synthetic information. 

Both parties agree that it was Leslie Wims Morris, head of corporate development for JPMorgan Chase’s consumer and community banking business, who asked Javice for more data in August 2021. Wims Morris made an unambiguous request for a list of Frank’s customer accounts, JPMorgan said, while Javice claims the exec made a late-stage request for Frank’s user data. 

Javice, in her answers, said that Wims Morris wanted Frank’s user list so the bank could verify its thesis that Frank would help it access the young student demographic. Frank, at this time, had already signed a letter of intent to sell to JPMorgan, but the transaction wasn’t completed. The bank was nearing the end of conducting diligence, according to Javice’s answers to the JPMorgan lawsuit.

Javice first pushed back on the request, and argued she couldn’t share the list owing to privacy issues, JPMorgan said in its lawsuit. Frank’s outside attorneys discussed the request on the phone with JPMorgan representatives and the bank’s legal counsel, Javice said. JPMorgan was advised by Dechert while Sidley Austin provided legal guidance to Frank, according to Javice’s answers. It was Frank’s outside attorneys who explained to JPMorgan why the request could not be fulfilled as initially asked, claiming that it could violate numerous privacy laws and federal regulations that limit the permissible use and sharing of FAFSA application data as well as personally identifiable information, according to the Javice court filing.

JPMorgan, Frank executives, and their representatives considered alternative options. Javice said she suggested providing the bank with temporary “read-only” access credentials to Frank’s accounts so they could view and validate user information directly, according to Javice’s court filing. Wims Morris declined the offer and instead asked Frank to provide synthetic data on the 4.26 million users to third-party vendor Acxiom, Javice claimed in her answers. JPMorgan had used Acxiom frequently for similar purposes, she said. The company was charged with validating and analyzing the data and then reporting back its conclusions to JPMorgan’s diligence team, according to Javice’s responses. 

This doesn’t jibe with what JPMorgan Chase says happened. Nowhere in its lawsuit does the bank say it agreed to receive synthetic data, according to the Dec. 22 court filing. JPMorgan asked for a list of Frank’s customer accounts and thought that’s what it received, the bank’s lawsuit said. When it agreed to the Frank acquisition, JPMorgan Chase said in court documents, it had no knowledge that the fake customer list Frank provided to Acxiom consisted of synthetic data. The bank said it had no reason to want synthetic data that mimics false data: “Synthetic data that mimics false data is of no use whatsoever as diligence material or, later, as marketing material.”

As a way to allay Javice’s fears, JPMorgan did agree to have Acxiom validate Frank’s customer information, rather than having the startup provide personal identifying information directly to JPMorgan, the bank’s court filing said. The bank believed that this would permit Frank to provide its list of actual customer accounts to Acxiom, with the data fields JPMorgan wanted, while protecting the personal information of Frank customer accounts and avoiding privacy issues, according to the JPMorgan lawsuit.

At this point, Frank engaged the services of the unnamed data science professor to conduct the “synthetic data project” that resulted in the list of more than 4 million users, Javice said. The Frank founder doubled the professor’s hourly rate to $600; the professor signed a Frank nondisclosure agreement on Aug. 3, JPMorgan said in its lawsuit. JPMorgan learned about the involvement of the data professor during an internal review later in 2022, according to the lawsuit.

At JPMorgan’s insistence, Javice said, she completed the task “over a couple of days and nights.” The professor uploaded Frank’s "customer list”—which now contained the synthetic data—to Acxiom on the morning of Aug. 5, the bank said. Acxiom analyzed the data and provided its report to JPMorgan later that evening. Acxiom destroyed the underlying data as its contract required, Javice said. Three days later, on Aug. 8, JPMorgan agreed to buy Frank for $175 million. Acxiom declined comment.

Though laypeople may not be familiar with synthetic data, it has become more common in recent years; Gartner estimates that by 2024, 60% of the data used for the development of A.I. and analytics projects will be synthetically generated, as the Wall Street Journal reported. The tool is an accepted and known solution for sharing the characteristics of restricted data, Javice said in the filing. Businesses use synthetic data to train algorithms to seek out fraud, to help self-driving cars adjust to new situations, and to gain insight into customer habits. Theodore Claypoole, a partner at law firm Womble Bond Dickinson who leads the intellectual property transaction team, said companies use synthetic data to “improve IT systems and to build and test software without exposing real, personal data.” Synthetic data also removes the risk of hacks that would expose personal information to bad actors. And finally, when an acquirer is checking out a target, sharing their biggest asset (a real customer list) could be a risky move if the deal doesn’t work out or if the acquirer is fishing for competitive intel.

JPMorgan Chase was familiar with synthetic data, Javice asserted in her answers, pointing to the fact that the bank has a dedicated team focused on the tool. The Frank founder provided a link to the JPMorgan synthetic data site, which said the method is used in trading, to prevent money laundering, and for fraud detection in payments.

Innocent until proven guilty

A major theme of Javice’s defense is that she did inform JPMorgan Chase about how many customers Frank had. The entrepreneur walked JPMorgan executives through the various stages where a Frank user would or wouldn’t submit personal info to the startup, while Frank performed product demonstrations, and even provided video recordings, according to Javice’s court document.

JPMorgan says in its lawsuit that Frank represented it had 4.25 million customers. The entrepreneur claims this is inaccurate. She provided a chart from a July 2021 management presentation showing how Frank had grown to over 4.25 million cumulative users since its first product launch in 2017, the Javice filing said. (The entrepreneur defined “user” as an individual who created a Frank account by entering a first name, last name, email, and phone number on Frank’s website, JPMorgan said in its lawsuit. Javice distinguished “users” from “website visitors,” representing to JPMorgan’s diligence team that since 2017 Frank had more than 35 million visitors to its site, JPMorgan’s lawsuit said). But the actual number of users who had filed FAFSA accounts with Frank was far smaller. Javice also mentioned an undated diligence meeting where a JPMorgan Chase executive asked her why the Frank website and press mentioned about 350,000 students, but the management presentation said 4.25 million students, according to her court filing. Javice said she explained that the 350,000 figure represented cumulative users who had filed FAFSA accounts with Frank, while the 4.25 million figure represented cumulative website users who viewed Frank content. That’s why the Frank website in 2021 could accurately state that it helped “more than 350,000 people access financial aid assistance,” while its Twitter posting said Frank had engaged “over 4 million students on how to pay for college," according to Javice's court documents. JPMorgan's lawsuit said it was seeking users, or customers, that it could market its products to, and had no use for website visitors.

Javice, whose guilt has been assumed by many in the media, also argued that Frank’s small marketing spend, around $2.25 million, should have clued in JPMorgan to how many users the startup had in 2021; Frank's $2.25 million figure was mentioned in diligence, in documents in the data room, and on its balance sheet, Javice said in the court document. Frank repeatedly indicated to the bank that the cost of acquiring a registered user of a Frank FAFSA account was around $5, according to the Javice filing. If Frank had 4.25 million users, then Frank’s marketing spend would come to more than $21 million, far more than the $2.25 million the startup cited. Even if JPMorgan Chase didn’t do the math, all the bank had to do was look at Frank’s diligence materials, which repeatedly referenced an increased number of roughly 500,000 students who filed a FAFSA with Frank, the court document said.

There’s also the relatively small price that JPMorgan Chase, a leading advisor on mergers, paid for Frank. The startup, when it sold in September 2021, was a seemingly high-flying fintech that had raised more than $20 million in funding. Frank had some big-name investors including Marc Rowan, cofounder and CEO of Apollo Global Management, an alternative asset manager; early-stage venture firm Aleph; and online education company Chegg. JPMorgan Chase scooped up Frank for $175 million, a low price for a company that claimed to have one-quarter of the student market. In September 2021, comparable companies at the time were trading at higher valuations. Chegg, which claims to have 36% of the student market, had a $10.9 billion market capitalization in September 2021. (Chegg’s valuation has since dropped about 81% to $2.1 billion.)

Wherever the legal case lands, it’s clear that a case of synthetic data created a real-world mess. 

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.