How Developers, Analysts, and Auditors Can Help Protect U.S. Investors from False Corporate Numbers
By Md. Tauhid Hossain Rubel, Doctoral Candidate and Researcher | Artificial Intelligence, Data Analytics, Cybersecurity & Financial Intelligence, United States
This is a short version of my longer Medium article. The full piece, with tables and all sources, is here https://mdtauhidhossainrubel.medium.com/fake-numbers-real-damage-01793b0acba2?sharedUserId=mdtauhidhossainrubel
Executive Summary
Fake financial statements are a quiet threat to the U.S. economy. They are false messages from trusted institutions, and they push investors, banks, and workers to act on a false picture. The ACFE 2024 report found that financial statement fraud is only 5% of fraud cases but has the highest median loss, at $766,000 (ACFE, 2024). New SEC cases from 2026 show the same pattern at a larger size. This post explains the problem in plain words, shows four simple data checks that anyone in data or finance can learn, and ends with controls that companies can add now. The people who gain are investors, auditors, regulators, and honest companies.
Keywords: Financial Statement Fraud; Data Analytics; Forensic Accounting; U.S. Economy; Investor Protection
Why a Developer or Analyst Should Care
Most of us use data to learn what is true. A financial report is data too. When someone fakes it, every model built on top of it is wrong, and so is every decision made from it.
The U.S. has deep markets and strong rules. Even so, the SEC keeps filing cases in which revenue was invented, bank statements were forged, or documents were made up to fool auditors. The SEC lists many of them as Accounting and Auditing Enforcement Releases. One example is release number 4583 from January 27, 2026, about a data company whose revenue was allegedly overstated by an average of 27% (SEC, 2026a).
I think of this as institutional disinformation. It has a false claim, a trusted source, and an audience that acts on it. The good news is that false numbers leave marks, and data work can find them.
The Numbers Behind the Problem
The ACFE studied 1,921 cases in 138 countries and territories, with about $3.1 billion in reported losses (Wipfli, 2024). Theft of assets was 89% of cases with a median loss of $120,000. Corruption was 48% with a median loss of $200,000. Financial statement fraud was only 5%, but the median loss was $766,000, up from $593,000 two years earlier (ASIS International, 2024; Selden Fox, 2024).
So the rare type is the costly type. That is a good reason to build checks for it.
Three Real Cases, Briefly
These are allegations in SEC filings. The people named can defend themselves in court.
Near Intelligence. The SEC says two companies sent inflated invoices to each other so one of them could book its own cash as revenue. Revenue was allegedly overstated by an average of 27% (SEC, 2026a).
Lugano Diamonds. The SEC says investor money was recorded as revenue and repayments were hidden as inventory buys. The complaint says over a billion dollars of fictitious revenue was recorded by Lugano and its public parent (SEC, 2026b).
GameOn. The SEC says a startup leader gave investors false financial statements and forged bank statements, though yearly revenue never passed $500,000. It also says he made fake email accounts that pretended to be bankers and advisers (SEC, n.d.).
Four Simple Checks You Can Learn
Check 1. Revenue Versus Cash
Real sales turn into cash. Compare revenue growth to cash from operations. If sales jump and cash stays flat, ask why. If cash does arrive, ask who sent it.
Check 2. Customer and Vendor Overlap
Match the customer list against the vendor list and bank accounts. A name that appears on both sides with large amounts is a flag. This is the pattern in round trip schemes.
Check 3. The Digit Test
In many real data sets, small first digits like 1 and 2 show up more often than large ones like 8 and 9. This is Benford Law. Made up numbers often do not follow it. It is not proof of fraud. It only tells you where to look first.
Here is a tiny example in Python. It counts first digits in a list of invoice amounts and prints the share of each.
from collections import Counter
def first_digit_share(amounts):
digits = [int(str(abs(a)).lstrip('0.').replace('.', '')[0]) for a in amounts if a != 0]
counts = Counter(digits)
total = sum(counts.values())
return {d: round(counts[d] / total, 3) for d in range(1, 10)}
Compare the output to the expected Benford shares. About 30% of numbers should start with 1 and about 5% with 9. Big gaps deserve a closer look.
Check 4. A Score Model
Messod Beneish (1999) built a model that uses ratios from financial statements, such as sales growth and changes in receivables, to estimate the chance that earnings were pushed up. It is a screening tool, not a verdict.
Do Controls Work
Yes. In the ACFE data, proactive data monitoring cut the median length of fraud by 58%, from 24 months to 10 months, and cut the median loss by 52%, from $200,000 to $96,000 (Wipfli, 2024). Surprise audits, outside audits, hotlines, and proactive data analysis were each tied to at least a 50% drop in loss and duration (Counter Fraud Centre Australia, n.d.).
What Teams Can Add This Month
First, ask the bank to send confirmations straight to the auditor or audit committee. Second, run the four checks above every month. Third, set up an outside hotline that reports to the audit committee. Fourth, confirm large deals by calling a known number, because fake emails are easy to make.
Why It Matters for the Country
Capital markets run on trust. When reports are fake, honest firms pay more to raise money, and households that hold stocks through retirement plans can lose savings. The U.S. needs more people who can combine finance and data skills. Those people protect investors directly. Open data from enforcement actions, linked to the red flags found, would help researchers build better tools.
Conclusion
Fake reports are not just accounting errors. They are false messages that spread harm. Simple data checks and a few strong controls can find them sooner. The earlier the discovery, the smaller the loss.
References
ACFE. (2024). Occupational fraud 2024. A report to the nations. https://legacy.acfe.com/report-to-the-nations/2024/
ASIS International. (2024). Beyond Bankman-Fried. https://www.asisonline.org/link/df59ad02cbe248c4b0b2b2a7f8b92832.aspx
Beneish, M. D. (1999). The detection of earnings manipulation. Financial Analysts Journal, 55(5), 24 to 36. https://doi.org/10.2469/faj.v55.n5.2296
Counter Fraud Centre Australia. (n.d.). Digest the Occupational Fraud Report 2024. https://www.counterfraud.gov.au/node/581
Securities and Exchange Commission. (n.d.). Litigation Release No. 26232. https://www.sec.gov/enforcement-litigation/litigation-releases/lr-26232
Securities and Exchange Commission. (2026a). Litigation Release No. 26469. https://www.sec.gov/enforcement-litigation/litigation-releases/lr-26469
Securities and Exchange Commission. (2026b). Litigation Release No. 26625. https://www.sec.gov/enforcement-litigation/litigation-releases/lr-26625
Selden Fox. (2024). 2024 ACFE report on occupational fraud. https://www.seldenfox.com/our-insights/articles/2024-acfe-report-occupational-fraud/
Wipfli. (2024). Fraud in focus. https://www.wipfli.com/insights/articles/ra-identifying-the-risks-of-occupational-fraud
Author Note
Md. Tauhid Hossain Rubel is a doctoral candidate and researcher specializing in artificial intelligence, data analytics, cybersecurity, and financial intelligence, with a focus on advancing U.S. economic resilience, investor protection, and technological leadership.
Top comments (1)
tr.ee/dev-to