AI Unemployment and Predatory Lending: A Copywriter's Double Squeeze

When we talk about AI's impact on employment, the discussion tends to stay at the level of abstract macro predictions. But a recent episode of Vox's podcast The Gray Area made that impact startlingly concrete — writer Daniel Kollitz recounted how he became part of "the first wave of professions displaced by AI," and how, after losing his income, he fell into a predatory lending trap disguised as a slick app.

This is a story about a "double squeeze": on one end, white-collar work being pushed out by AI; on the other, financial products specifically designed to exploit those who've been pushed out. Kollitz documented this experience in his piece Paid Phones for The Baffler, and his account on the podcast cuts deeper than any AI unemployment report.

The First Job AI Took From Him

Kollitz used to make a living writing SEO-optimized promotional web content. In his own words: "I used to make my living writing SEO-optimized promotional copy, and I thought the good times would never end."

The comfort of that life was striking. He described his daily routine: sleeping until 11, drifting through the morning, heading to his office — which he rented for $600 a month — checking his email for two, three, sometimes five assignments, often agreeing to them without reading the brief, knocking them out in an hour or two, and having the rest of the day to himself. A piece on "video surveillance as a service" could earn him $400 in 15 minutes.

"The pricing in this industry made no sense whatsoever, but that was great." He admitted the "wiggle room" let a lot of people live better lives than they might have otherwise. For him, it was never his real "work" — his true writing was literary nonfiction and fiction; the copywriting was just "what he did for money."

This semi-bohemian New York life lasted about five years. Then ChatGPT arrived, and it was over.

Why His Profession Was First in Line

Kollitz said that the moment ChatGPT launched, he knew "it was going to ruin my life." The reason was the extreme formulaic nature of what he wrote.

He broke down the fixed structure of these articles clearly: identify a problem, frame the product as the solution (without naming it directly, since it's disguised advertising), explain the historical background of the problem, detail why the solution works, wrap it up in a closing paragraph. "This stuff was so mechanical, so formulaic, that a machine designed to mass-produce repetitive, low-quality text could replace me in a second."

The displacement was gradual and wave-like. In the early days of the technology, there was almost no impact; when the model improved slightly, work contracted; then clients overcorrected and work flowed back; with each new model iteration, the cycle repeated — until the moment "the industry just died."

Kollitz's dark humor lay in his absurd status as a pioneer: "I cannot believe that this specific job of mine was the first thing destroyed by this supposedly omnipotent technology that's going to transform every profession and maybe kill us all." He noted that AI awareness had spread across society in wildly uneven ways — some only encountered it as news, some were obsessed with doomsday scenarios, some saw their work trimmed slightly. Standing right at the center, he arrived at a deeply pessimistic view of the technology.

---

SEO (Search Engine Optimization) copywriting is content written specifically to improve a webpage's ranking on search engines like Google. The core logic: search algorithms tend to favor pages containing specific keywords and following certain structural formats, which gave rise to a massive industry producing articles from fixed templates — identify target keywords, build a problem-solution framework, fill in background information, insert soft brand advertising — all highly formulaic. This content demands no creativity or deep insight at the language level, only fluency and format compliance, which put it squarely in the sweet spot of early large language models. Kollitz's case illustrates a basic principle of AI displacement: the first jobs to go aren't necessarily "low-skill" ones, but rather "highly structured jobs that can be decomposed into predictable steps" — regardless of how much education those steps previously required.

---

Predatory Lending Dressed Up as a Friend

After losing his income, Kollitz fell behind on his credit cards and soon met "the gang" — Albert, Cleo, Bridget, Dave, and a roster of other payday loan apps. Their names were deliberately humanized; Dave's mascot was an adorable little bear. They were "shiny, millennial-coded, sitting on your iPhone right next to Instagram without looking out of place."

But their essence was no different from the grimy payday loan storefronts of earlier eras. Kollitz walked through Albert's fee structure from personal experience: to get approved for a $25 loan, he first had to pay $20 to subscribe to the "Albert Standard" plan; because he needed the money urgently and couldn't wait three or four days for the free transfer, he paid another $7 for instant delivery. The result: to receive a $25 loan, he paid $27 upfront.

Even more alarming is how these apps make money. Kollitz cited an FTC action under Lena Khan during the late Biden administration that sued companies in this space and obtained roughly $18 million in settlements from Bridget and Cleo. The investigation revealed that most users were never actually approved for loans — yet they were still charged monthly subscription fees of $5 to $10. They paid, and got nothing in return.

These apps also used AI on two levels: first, to analyze users' bank transaction histories to gauge how much repayment capacity could be squeezed out of them; second, apps like Cleo came with AI chatbots capable of delivering financial advice with a "cool attitude" — Cleo even launched a "roast your ex" feature for Valentine's Day. Cleo reportedly hired 15 human comedy writers to craft jokes mocking users' financial situations — Kollitz remarked sharply that those writers "should be ashamed of themselves." He described the scene as medieval: a desperate serf begging for a coin, while the court jester in the corner mocks him.

The lineage of these apps is also telling: Cleo's founder previously worked at the notorious British payday lender Wonga (which was caught fabricating fake law firms to chase debts), and Bridget was acquired by the debt-holding group that owns Rent-a-Center.

---

Payday loans are short-term, small-dollar loans due by the borrower's next payday, with a history spanning decades in the United States. The traditional form was the brick-and-mortar storefront, lending to low-income people in urgent need of cash at extremely high annual percentage rates (APR often reaching 300%–400%). After the 2010s, this model migrated en masse to mobile, evolving into "cash advance apps." This new generation deliberately obscures its fee structure — wrapping interest as "membership subscription fees," labeling transaction fees as "instant transfer tips," and lowering users' guard with cute brand identities and social-media-style interfaces. The CFPB and FTC have investigated these apps multiple times, finding that they remain payday loans in substance, just wearing a digital veneer. The enforcement actions taken against this industry during Lena Khan's FTC tenure were among the more substantive consumer protection outcomes of the Biden administration.

---

Reddit Communities and the Collective Resistance of ACH Revocation

The turning point in the story was a box of Milk Duds he couldn't afford. Kollitz had opened a Venmo card and stashed $60 on it to keep the apps from pulling directly from his bank account. But at the supermarket, the Venmo card wasn't accepted. He ran outside, transferred $4 into his bank account, and jogged back to the register — within just 10 seconds, Cleo or Albert had already swept it away. A 33-year-old man who couldn't afford a box of candy. "Pretty grim," he said.

That night, he discovered a Reddit community for "cash advance apps" — membership had grown from 20,000–30,000 to over 60,000. Two things were happening there simultaneously: a political awakening, with people beginning to discuss how the power elite and the "Epstein class" were bleeding ordinary people dry; and the spread of a technical workaround called ACH revocation.

To sidestep regulations, these apps had structured their loans as "non-recourse advances," meaning they couldn't hand the debt to a collections agency. Users could send the company written notice revoking its access to their bank accounts. Kollitz watched this method "go from rumor to action plan" — today that subreddit has an ACH revocation guide pinned at the top. The trade-off is that once you revoke, you can no longer use these apps.

Kollitz never tried it himself — partly because he still wanted access to the apps, and partly because he found it hard to believe there was such a thing as a free lunch.

---

ACH (Automated Clearing House) is the core infrastructure for interbank electronic transfers in the United States, handling direct deposits, automatic bill payments, and other batch transactions. When users sign up for these apps, they typically authorize the app to debit their bank account via the ACH system. Under the federal Electronic Fund Transfer Act (EFTA) and Nacha rules (the governing body of the ACH network), consumers have the right to revoke a third party's ACH debit authorization at any time in writing — by sending written notice to the lender and instructing their bank to reject debits from that party. Because these cash advance apps structured their loans as "non-recourse advances," once a user revokes ACH authorization and the app can no longer pull funds, they have no ability to sell the debt to collectors and no legal recourse — effectively creating a regulatory arbitrage loophole that user communities discovered and began exploiting.

---

Two Futures: Collapse or Barely Getting By

The deepest insight in the piece is this: these payday loan apps don't actually want mass unemployment. Their ideal customer is someone "nearly broke but still pulling in some income" — someone who keeps coming back and repays at least part of what they owe.

When asked which future was more frightening — a full AI-driven economic collapse with mass unemployment, or more people like him scraping by "loan by loan" — Kollitz said the latter was "more likely, and more consistent with where this country currently is." A total apocalyptic collapse might spark revolutionary change, but it could just as easily lead to fascist crackdowns.

He also honestly acknowledged his own position: as someone from an upper-middle-class family who could always move back to his parents' house, he was not among the worst-off victims of this economy. And as his finances began to recover after another piece went viral, he half-jokingly called himself a "class traitor" to the Reddit payday loan community.

Kollitz sees this experience as continuous with his earlier writing about short-form porn eroding attention spans: the kind of life he valued — light work plus meaningful creative output — is being squeezed from both ends. There's no room left to make a decent living with easy work while pursuing real creative projects, and no audience willing to stop long enough to engage with that art. "Any breathing room outside of working ten hours a day is disappearing."

Perhaps this is the most honest picture of the AI economy: not a science-fiction apocalypse, but countless people being slowly ground down by small debts, one loan at a time, beneath the candy-coated surface of a slick app.