By Futurist Thomas Frey
Banking used to be straightforward: you deposited money, the bank paid you interest, they lent it to others at higher rates, everyone understood the deal. That simplicity died decades ago, replaced by a baroque system of fees, penalties, and charges so complex that even bank employees often can’t explain them.
AI is now analyzing millions of customer accounts, transactions, and fee schedules. What it’s revealing is a systematic extraction economy—a financial system that has quietly evolved to profit from confusion, from mistakes, from the poorest customers, and from complexity that serves no purpose except generating revenue.
The awakening in financial services isn’t just about unfair fees. It’s about revealing an entire industry that restructured itself to profit from customer disadvantage while maintaining the appearance of serving customer interests.
The Overdraft Profit Machine
Overdraft fees are supposedly designed to discourage customers from spending money they don’t have. In practice, AI analysis reveals they’re designed to maximize fee extraction from customers living paycheck to paycheck.
Here’s how it works: Banks process transactions in whatever order generates the most fees. You might make several small purchases throughout the day, then one large payment. If the bank processes the large payment first, multiple small transactions overdraft, generating multiple $35 fees instead of one.
AI analysis of millions of transaction processing patterns reveals this isn’t random—it’s systematic. Banks that claim to process transactions “as they occur” actually process largest-to-smallest to maximize overdraft fees. One analysis found that simply processing transactions chronologically would reduce overdraft fees by approximately 50%. Banks know this. They do it anyway.
Even more predatory: AI has revealed the practice of “phantom” overdrafts. A customer with $100 in their account makes a $90 purchase. The bank immediately places a hold for the $90, showing available balance of $10. The customer then makes an $8 purchase. The bank approves it—but processes the transactions in an order that generates an overdraft fee, even though the customer never actually had insufficient funds at the time of either transaction.
Analysis shows that approximately 9% of bank customers generate roughly 84% of overdraft fee revenue. These aren’t wealthy people carelessly managing large accounts. They’re low-income customers living on the edge, for whom a single $35 overdraft fee might mean choosing between medication and groceries.
The most damning finding: banks’ own internal analysis shows they could eliminate most overdrafts with a simple real-time balance notification system. Such systems cost virtually nothing to implement—most banks already have the technology. They don’t implement it because overdraft fees generate approximately $15 billion annually across the industry. The fees aren’t a byproduct of banking—they’re a business model.
The Minimum Balance Trap
Banks claim that minimum balance requirements ensure customers maintain healthy financial habits. AI analysis tells a different story: minimum balance fees are a poverty tax that extracts wealth from those who can least afford it.
Here’s the pattern AI revealed: banks set minimum balances just high enough to be difficult for lower-income customers to maintain, but low enough that middle and upper-income customers never notice them. A checking account requiring a $1,500 minimum balance costs a wealthy customer nothing—they always have $1,500. But a customer living paycheck to paycheck might dip below $1,500 briefly each month, triggering a $12-15 fee.
AI analysis of account patterns shows that customers who pay minimum balance fees average 8-10 fees per year—nearly $120-150 extracted annually from people who can least afford it. Over a decade, that’s $1,200-1,500 taken from low-income customers while wealthier customers using identical services pay nothing.
Even more troubling: AI has revealed that banks systematically steer low-income customers toward accounts with monthly fees while offering wealthy customers free accounts with better terms. The exact same services—checking, savings, online banking—but priced inversely to ability to pay.
One analysis found that customers in lower-income zip codes paid an average of $240 per year in various bank fees, while customers in wealthy zip codes paid an average of $18 per year for objectively superior banking services. The system literally charges poor people more for worse service.
The Credit Card Interest Shuffle
Credit card interest rates are supposedly based on risk—riskier customers pay higher rates. But AI analysis of credit card pricing reveals something far more complicated and far less justified.
First, the rate variation is staggering. AI has documented that customers with identical credit scores, income levels, and debt-to-income ratios often receive dramatically different interest rate offers from the same bank—variations of 5-10 percentage points. The pricing isn’t based on risk modeling; it’s based on what the bank thinks it can extract.
More specifically, AI has revealed that credit card companies systematically test price sensitivity. They offer different rates to similar customers to see who accepts. Those who accept higher rates get flagged as “price insensitive” and receive worse offers going forward. The system is explicitly designed to identify and exploit customers who don’t shop around.
Even more problematic: the “teaser rate” pattern. AI analysis shows that approximately 65% of customers who accept low introductory rates on balance transfers never successfully pay off the balance before the rate jumps. The banks know this—their internal models predict it with remarkable accuracy. The teaser rate isn’t designed to help customers escape debt; it’s designed to transfer balances to the offering bank, then extract maximum interest when rates reset.
AI has also revealed the practice of “rate jacking”—raising interest rates on existing balances based on factors unrelated to the cardholder’s behavior with that specific card. A customer with perfect payment history on their Bank A card can see their rate increase because they missed a payment to Bank B, or because their credit utilization increased temporarily, or because they closed an unused account (which “hurt their credit mix”).
These rate increases apply to existing balances—debt incurred at the promised rate suddenly becomes subject to penalty rates of 25-30%, sometimes on tens of thousands of dollars. AI analysis shows this practice generates approximately $8-10 billion annually in additional interest payments that bear no relationship to actual credit risk.
The Savings Account Illusion
Banks advertise savings accounts as places to safely store money and earn interest. AI analysis of savings account rates versus bank investment returns reveals the reality: savings accounts are where banks acquire capital virtually free.
The average savings account in America pays approximately 0.01-0.4% interest (traditional banks). Meanwhile, those same banks invest deposits at returns of 4-8%. The spread—the difference between what they pay depositors and what they earn investing deposits—represents pure profit.
AI analysis shows this spread has widened dramatically over the past two decades. In the 1980s, savings accounts paid roughly 70-80% of prevailing interest rates. Today, they pay roughly 5-10%. When the Federal Reserve raises rates to combat inflation, banks immediately raise loan rates but delay raising deposit rates by months or longer—capturing the spread.
Even more specifically, AI has revealed that banks segment customers and pay dramatically different rates. Customers with large balances who might shop around get offered higher-yield savings products. Customers with small balances get the default 0.01% accounts. The same institution, the same FDIC insurance, but returns that vary by 50-100x based on balance size and likelihood of switching banks.
One analysis found that if all savings account holders switched to high-yield online savings accounts (currently offering 4-5%), Americans would collectively earn approximately $150-200 billion more per year in interest. Traditional banks rely on customer inertia and lack of awareness to maintain the profit extraction.
The Payment Processing Markup
Every time you swipe a credit or debit card, the merchant pays a fee—typically 1.5-3.5% of the transaction. Consumers rarely see this directly, but they pay it through higher prices. AI analysis of payment processing reveals an industry built on opacity and monopoly power.
Here’s what AI discovered: the payment processing system involves multiple layers—card networks (Visa, Mastercard), issuing banks, acquiring banks, and payment processors—each taking a cut. The total fees often exceed the actual cost of processing the transaction by 300-500%.
The real cost of electronically moving money from one account to another? Fractions of a penny. Yet a $100 transaction might generate $2-3.50 in fees. AI analysis estimates that payment processing generates approximately $100 billion in fees annually in the U.S. alone—nearly all of which represents markup over actual costs.
Even more troubling: the fee structure is deliberately opaque. Merchants receive statements with dozens of line items, abbreviations, and codes that make it nearly impossible to understand what they’re paying for. AI analysis of merchant processing statements found that approximately 30-40% of merchants are paying for services they don’t use or paying rates higher than their contracts specify—and don’t know it because the statements are intentionally incomprehensible.
The system also extracts disproportionately from small businesses. Large retailers negotiate rates of 1.5-2%, while small businesses pay 2.5-3.5% for identical services. The mom-and-pop restaurant pays double the processing rate of the national chain—purely because of negotiating power.
The Investment Fee Compound Problem
Investment fees seem small—1% or 2% annually doesn’t sound like much. But AI analysis of long-term investment returns reveals that these “small” fees compound into staggering amounts over decades.
A simple example: $100,000 invested at 8% annual return becomes $1,006,266 after 30 years. But with a 2% annual fee, that same $100,000 becomes $574,349. The fee doesn’t take 2%—it takes 43% of your total wealth over three decades through compounding.
AI analysis of actual investment performance data reveals something even more troubling: actively managed funds charging these higher fees consistently underperform low-cost index funds after fees. Approximately 85-90% of actively managed funds fail to beat their benchmark indexes over 15-year periods. Investors pay extra for worse results.
Even more specifically, AI has revealed that many investment advisors are paid based on assets under management rather than performance. This creates perverse incentives: advisors do better when clients save more, but don’t suffer when clients lose money through poor advice. One analysis found that during market downturns, advisors’ fees remain steady even as client portfolios collapse—the risk is entirely one-sided.
The 401(k) system reveals this problem at scale. AI analysis of 401(k) fee structures shows that many employer-sponsored retirement plans include layers of fees that participants don’t see or understand: fund management fees, plan administrative fees, record-keeping fees, and sometimes hidden revenue-sharing arrangements. These fees can total 1.5-2.5% annually, dramatically reducing retirement wealth.
One comprehensive analysis estimated that a median-income worker who saves diligently throughout their career will pay approximately $138,000 in 401(k) fees over their lifetime. That’s money that could have compounded for retirement instead of funding the financial services industry.
The Mortgage Origination Markup
Getting a mortgage involves paying thousands of dollars in “origination fees,” “processing fees,” “underwriting fees,” and various other charges. AI analysis of mortgage costs reveals that many of these fees bear no relationship to actual costs incurred.
By analyzing mortgage closing costs across thousands of transactions, AI has identified that identical mortgages from the same lender can have closing costs that vary by 40-60% based primarily on whether the borrower negotiates or accepts the initial fee schedule. The fees aren’t based on cost—they’re based on what the lender thinks they can charge.
Even more specifically, AI has revealed systematic overcharging on title insurance, appraisals, and credit reports. The title insurance that costs $1,500 in one state costs $400 in another for identical properties and identical risk. The appraisal that costs $500 could be done equally well for $200. Credit reports that cost lenders $8-12 get charged to borrowers at $30-50 each.
One analysis estimated that opaque mortgage fees extract approximately $10-15 billion annually beyond actual costs. And because mortgages are complex, emotionally fraught transactions with tight timelines, borrowers rarely question the fees or shop around effectively.
AI has also exposed the practice of “yield spread premiums”—hidden kickbacks to mortgage brokers for steering borrowers into higher-interest loans than they qualify for. While technically disclosed, these payments are buried in closing documents that most borrowers don’t understand. The broker presents themselves as working for the customer while actually being paid to work against the customer’s interests.
The Foreign Transaction Fee Scam
When you use your credit or debit card abroad, you’re typically charged a “foreign transaction fee” of 2-3%. Banks claim this covers currency conversion costs and international processing.
AI analysis of actual currency conversion costs reveals the truth: converting currency costs fractions of a percentage point. The 2-3% fee is nearly pure markup. Even more egregiously, some transactions that occur entirely in U.S. dollars still incur foreign transaction fees simply because the merchant is located abroad.
What really exposed the scam: many banks now offer credit cards with no foreign transaction fees—proving conclusively that the fees aren’t necessary. They’re simply extraction from customers who don’t know better or don’t have alternatives.
AI analysis estimates that foreign transaction fees generate approximately $4-5 billion annually for U.S. banks—nearly all of which represents profit beyond actual costs.
The Wire Transfer Anachronism
Sending a domestic wire transfer typically costs $25-35. Receiving one costs $10-20. Banks claim these fees reflect the cost of secure, immediate money transfers.
But AI analysis reveals that wire transfers use technology from the 1970s that costs virtually nothing to operate at scale. The actual cost per transaction: pennies. Meanwhile, apps like Venmo, Zelle, and Cash App transfer money instantly for free.
So why do banks charge $25-35 for wires? Because they can. Business customers and people in urgent situations will pay it. AI analysis shows that wire transfer fees generate approximately $2-3 billion annually for banks—representing roughly 98-99% profit margin.
The same pattern holds for international wires, which can cost $40-50 and take 3-5 days despite occurring electronically. Services like Wise (formerly TransferWise) now do international transfers in hours for 1/10th the cost, proving that traditional bank fees bear no relationship to actual costs.
The Wealth Management Two-Tier System
Banks market themselves as serving everyone’s financial needs equally. AI analysis of actual bank behavior reveals a starkly different reality: banks have essentially created two separate systems—one for wealthy clients, one for everyone else.
AI has analyzed the distribution of bank services and found that customers with investable assets above $250,000 receive:
- Fee waivers on virtually all accounts
- Dedicated relationship managers
- Better interest rates on deposits
- Lower interest rates on loans
- Faster service and dispute resolution
- Access to investment products with lower fees
Customers below that threshold receive:
- Fee-laden accounts
- Automated service with long wait times
- Market-rate or worse deposit rates
- Higher loan rates despite identical credit profiles
- Slower dispute resolution
- Access only to high-fee investment products
The same bank, but effectively two different institutions based on account size. AI analysis shows that this segmentation has intensified over the past twenty years as banks realized that serving lower-income customers was less profitable than extracting fees from them.
The ATM Fee Multiplication
ATM fees represent a particularly clear example of the extraction economy. You pay your bank a fee to access your own money. The ATM owner charges you another fee. Both fees have risen steadily despite declining costs.
AI analysis of ATM fees reveals several troubling patterns. First, fees are highest in low-income neighborhoods—the exact places where people can least afford them. A low-income person accessing cash in their own neighborhood might pay $3-5 in combined fees, while someone in a wealthy suburb pays nothing at their bank’s free ATM.
Second, banks have systematically reduced the number of fee-free ATMs while raising out-of-network ATM fees, essentially forcing more customers to pay fees. One analysis found that the ratio of bank customers to fee-free ATMs has increased by approximately 60% over fifteen years—a deliberate strategy to generate fee revenue.
Third, the fees bear no relationship to costs. Operating an ATM costs roughly $0.05-0.10 per transaction in variable costs. The $2-3 fee represents 2,000-3,000% markup.
AI estimates that ATM fees extract approximately $8-10 billion annually from customers, with low-income customers paying disproportionately.
The Branch Closure Strategy
Banks have closed tens of thousands of branches over the past decade, claiming customers prefer online banking. AI analysis of branch closure patterns reveals a more troubling story.
Branches in wealthy areas remain open. Branches in low-income and rural areas close at 3-4 times the rate. The effect: people with means have convenient access to in-person banking, while those without reliable internet access, those who need in-person services, and those in rural areas face increasing difficulty accessing basic banking.
AI has revealed that branch closures correlate strongly with neighborhood income rather than with profitability or customer preference. Banks are systematically withdrawing from communities they view as less profitable, creating banking deserts where residents must travel significant distances for in-person service.
This matters particularly for elderly customers, recent immigrants, and people without reliable internet access who depend on branch banking. The digital divide becomes a banking access divide.
The Coming Reckoning
The financial services industry has operated on information asymmetry for decades. Customers couldn’t easily compare fees, couldn’t see how much they were paying relative to actual costs, couldn’t identify that they were receiving different treatment based on wealth.
AI eliminates that asymmetry. Apps now exist that analyze your banking fees and suggest cheaper alternatives. Comparison tools reveal exactly how much more you’re paying than necessary. Pattern analysis shows which fees are justified by costs and which are pure extraction.
The industry response has been predictable: lobbying against transparency requirements, adding arbitration clauses that prevent class action lawsuits, and creating complexity fast enough to stay ahead of consumer tools.
But the fundamental economics are against them. Once customers can see that they’re paying 100x more than necessary for basic services, once the extraction becomes visible and quantifiable, the system becomes indefensible.
We’re already seeing the early stages of disruption. Fintech companies offering checking accounts with no fees, no minimums, and better rates. Digital banks with transparent fee structures. Peer-to-peer payment systems that bypass traditional banking entirely.
Traditional banks have two choices: reform toward genuine value creation and transparent pricing, or double down on extraction while they still can. The early evidence suggests most are choosing the latter, which means the disruption, when it fully arrives, will be more severe.
Final Thoughts
The awakening in financial services isn’t about demonizing banks. Many banking functions are genuinely valuable—safeguarding deposits, facilitating commerce, allocating capital efficiently. The problem is that over decades, the industry drifted from “how do we serve customers and make a reasonable profit” to “how much can we extract without customers noticing or leaving.”
AI is making them notice. And increasingly, making them leave.
The financial services industry could have reformed itself when the extraction became obvious internally. Instead, it chose to optimize the extraction while maintaining the appearance of customer service. Now AI is revealing the gap between appearance and reality at a scale that can’t be dismissed or explained away.
The transformation won’t happen overnight. But it’s coming. Every fee that can’t be justified by actual costs, every service that charges 10x-100x the underlying expense, every practice that extracts disproportionately from those who can least afford it—all of it is now visible, quantifiable, and increasingly indefensible.
The age of extraction through opacity is ending. What replaces it will depend on whether the industry embraces transparency or whether disruption forces it upon them.
In our next column: Insurance—The Bet You Can’t Win.
Related Articles:
Consumer Financial Protection Bureau – Data Spotlight: Overdraft/NSF Fee Reliance Since 2015
Pew Charitable Trusts – Overdrawn: Consumer Experiences With Overdraft
Financial Times – The $100bn Fee Bonanza Driving Wall Street’s Retail Banking Push
