Skip to content
Pandemic Darlings The pandemic economy, in original documents
Home Interactives Pandemic fraud, five years on
Pandemic Darlings · The Archive in Charts

Pandemic fraud, five years on: 2,709 federal cases, $4.1 billion charged in resolved cases

What 2,709 federal pandemic-fraud cases actually look like once you put them in one table: how big the frauds were, where they were charged, who the government went after, and how it all ends. Built from the Justice Department's own press releases.
2,709
DOJ enforcement actions
Mar 2020 – May 2026
~$2.1T
relief disbursed
SBA (PPP, EIDL) + DOL (UI)
>$200B
potential fraud, PPP + EIDL
SBA OIG estimate, June 2023
~0.2%
charged in resolved cases
$4.1B of the $2.1T disbursed
~2,005
convictions to Dec 31, 2023
1,045 PPP/EIDL (Mar 2025)
36 mo
median prison sentence
784 court-imposed terms
FL · 251
busiest state
of 50 + DC
1,540
wire-fraud charges
the workhorse

On a narrow screen, scroll each chart sideways to see all of it.

Build note: 4 of the 14 static charts were rendered on 2026-06-19, three on 2026-09-24, three on 2026-09-30 and four on 2026-10-09, when the case table, its tiles and the charts drawn from it were rebuilt from data release v18; the table, the filters and the tiles are built from the 2,709 releases this page carries. A chart can therefore run a few releases behind the table: the sentencing charts count 785 sentenced defendants, where this page’s data gives 791 sentencing releases. The case-status chart was redrawn on 2026-09-24 from this page’s data (699 charged, 791 sentenced).

Start with the number everyone repeats: more than $200 billion of pandemic business loans was “potentially fraudulent,” per the SBA’s own Inspector General, plus $100–135 billion of unemployment fraud per the GAO. Those are real estimates — but they are built from statistical indicators (duplicate addresses, reused IP addresses, dead Social Security numbers), not courtroom findings. Set them next to what the prosecutions actually account for: about 2,005 federal convictions (guilty pleas and trial verdicts, as of December 31, 2023), and $4.1 billion of charged dollars across the resolved cases whose DOJ release states a figure you can check against the release itself. Convictions undercount real fraud — most of it will never be prosecuted, and the ten-year filing window runs into the 2030s. But the distance between “a computer flagged this loan” and “a jury convicted this person” is the distance between an estimate and a fact.

Flagged is not the same as proven

SBA's Inspector General estimates that more than $200B of PPP and COVID-EIDL money was potentially fraudulent (June 2023); GAO puts unemployment-insurance fraud at $100–135B (September 2023). The two measure different programs in different ways, so they sit on separate bars and are not added. Cases that ended in a plea or verdict account for $4.1B of charged dollars, the full amount charged and an upper bound on what any court found was lost. Cases whose release states no checkable figure still count as cases here; they just carry no dollars, which makes the $4.1B a floor.

2026-10-09T00:00:00 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ $0B $50B $100B $150B $200B $250B SBA Inspector General: potential fraud, PPP + EIDL GAO: estimated fraud, unemployment insurance Resolved cases: charged dollars PPP $64B COVID-EIDL more than $136B more than $200B SBA OIG, June 2023 unemployment insurance $100–135B, a range GAO, Sept 2023 $4.1B charged in cases that ended in a plea or verdict DOJ releases, Mar 2020 – May 2026 The headline pandemic-fraud numbers are statistical estimates of fraudulent or “potentially fraudulent” dollars, not findings against any person. Two watchdogs measured different programs on different dates, so their estimates sit on separate bars and are not added. Flagged is not the same as proven Estimates: SBA OIG Report 23-09 (more than $200B in potentially fraudulent COVID-19 EIDLs and PPP loans) and GAO-23-106696 ($100–135B in unemployment-insurance fraud). A flag is not a conviction. Resolved-case dollars: Pandemic Darlings analysis of DOJ press releases — the amount charged in the 799 distinct schemes that ended in a plea or verdict AND carry a dollar figure stated in the release. Cases whose release states no verifiable figure are excluded, so this is a floor; it is also the amount charged, an upper bound on what any court found was lost.

Even the estimates disagree — and EIDL leaked worst

SBA's watchdog flags about 8% of PPP dollars as potentially fraudulent but 33% of COVID-EIDL — the program the government handed out directly, without the fintech middlemen that took the blame for PPP. The GAO puts unemployment fraud at 11–15%.

2026-09-24 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ 0% 10% 20% 30% 40% share of the program's money paid out Paycheck Protection (PPP) COVID-EIDL Pandemic unemployment (UI) 8% · $64B · SBA OIG, June 2023 33% · more than $136B · SBA OIG, June 2023 11–15% · $100–135B · GAO, Sept 2023 Gold = the share of each program's money its watchdog estimates was fraudulent or “potentially fraudulent,” in the watchdog's own percentages. The direct-to-borrower EIDL program flagged far higher than fintech-heavy PPP. The three estimates are not added together. Even the official estimates disagree — and EIDL leaked worst estimate upper end of GAO's range SBA OIG Report 23-09 (June 27, 2023): $64B is 8 percent of PPP disbursed funds; more than $136B is 33 percent of COVID-EIDL disbursed funds. GAO-23-106696 (Sept 2023): $100–135B is about 11 and 15 percent of the unemployment-insurance benefits paid during the pandemic.

And where there is real fraud, it should be trivial to prove

PPP and EIDL fraud is a documentary crime: a federal form signed under penalty of law against the government's own records. The same data matches that built the $200B estimate — dead Social Security numbers, EINs registered after the pandemic began, no IRS payroll behind a claimed payroll — are the evidence. The low conviction count is about prosecutorial capacity and small-dollar triage, not difficulty of proof.

2026-06-19T19:33:53.385060 image/svg+xml Matplotlib v3.10.9, https://matplotlib.org/ PPP and EIDL fraud is a documentary crime Every red flag the watchdogs used to ESTIMATE fraud is a record the government already holds to PROVE it. THE RED FLAG (what makes a loan “potentially fraudulent”) THE RECORD THAT SETTLES IT THE CHARGE Applicant's SSN belongs to a dead person SSA Death Master File Aggravated ID theft · §1028A Business's EIN was issued after the pandemic began IRS EIN registration date False statement · §1001 Payroll and employees were invented IRS Form 941 / W-2 filings (none) Wire fraud · §1343 2019 revenue was inflated to size the loan Prior-year IRS tax return False stmt. to a bank · §1014 One applicant, many loans SBA's own loan database (dup. EIN/bank/address) Wire fraud · conspiracy Funds bought a Lamborghini, not payroll Bank records (subpoena) Money laundering · §1957 The signal that flags the loan is most of the case. These are the same data matches behind the $200B “potentially fraudulent” estimate — and every PPP/EIDL applicant signed, under penalty of law (18 U.S.C. §§1001 & 1014), that the numbers were true. Indicators per SBA-OIG Report 23-09 and GAO-23-105331 (fraud schemes & indicators); certifications per SBA Form 2483. Fair caveat: a flag marks a loan to check. Some flagged loans turn out to be error or ineligibility, and what limits prosecution is capacity and small-dollar triage. Pandemic Darlings.

…and in people as well as dollars

About 1,045 PPP/EIDL fraud convictions (SBA OIG, March 2025), and about 2,005 across all pandemic programs (DOJ, as of December 31, 2023), against more than 15 million SBA loans. Convictions are a floor: most fraud is never charged, and prosecutors have until the 2030s.

2026-09-24 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ 11.5M PPP loans approved 3.68M COVID-EIDL loans ~1,045 PPP/EIDL fraud convictions ~2,005 DOJ pandemic convictions, to Dec 2023 Convictions are a floor, not a full count — most fraud is never prosecuted, and the 10-year filing window runs into the 2030s. …and in people: about 1 conviction for every ~15,000 SBA loans SBA (loan counts); SBA-OIG Semiannual Report to Congress, Mar 31 2025 (1,045 convictions); DOJ CFETF 2024 report (about 2,005 pleaded guilty or convicted at trial, as of Dec 31, 2023).

Pandemic fraud was a retail crime

Forget the $2 billion headlines. The median scheme the Justice Department charged is $1.2 million — one fake business, a loan or two. Four in ten come in under $1 million, and the ten biggest schemes carry nearly a quarter of all the dollars between them.

2026-10-09T00:00:00 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ $1k–10k $10k–100k $100k–1M $1M–10M $10M–100M $100M–1B 0 100 200 300 400 500 600 700 Number of schemes 3 157 453 666 156 12 DOJ schemes by dollar amount charged. Most are a single loan, not a syndicate. 1,447 distinct schemes with a verified dollar figure 42% are under $1 million (median $1.2M) the 10 biggest schemes = 23% of all the dollars Pandemic fraud was a retail crime Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 releases). Pandemic Darlings analysis. A scheme carries a dollar figure only where DOJ stated one in the headline, or where the case table’s figure matches one in the release’s opening paragraph. Program totals, multi-case round-ups and uncorroborated figures are excluded; amounts as charged/alleged.

Where the cases landed

Florida leads, with 251 cases — more than California or New York. The map is part crime scene and part prosecutorial temperament: it shows where U.S. Attorneys brought cases, which tracks both where the fraud was and which offices went looking for it. Massachusetts and the Middle District of Florida punch far above their population.

2026-06-19T16:25:25.277983 image/svg+xml Matplotlib v3.10.9, https://matplotlib.org/ AK 6 ME 22 VT 1 NH 18 WA 62 ID 11 MT 13 ND 2 MN 17 WI 23 IL 67 MI 45 NY 175 MA 152 RI 31 OR 18 NV 33 WY 4 SD 3 IA 26 IN 12 OH 52 PA 70 NJ 109 CT 22 CA 173 UT 24 CO 38 NE 8 MO 59 KY 41 WV 65 VA 77 MD 77 DE 13 AZ 12 NM 6 KS 11 AR 11 TN 44 NC 95 SC 17 DC 30 OK 29 LA 99 MS 33 AL 19 GA 108 HI 6 TX 95 FL 251 DOJ pandemic-fraud actions by state. Darker = more cases. Where the cases landed SDWV SDNY WDNY NDGA EDVA EDLA MD CDCA SDFL NJ MDFL MA 54 56 57 58 68 70 77 87 106 109 129 152 Most active U.S. Attorney districts 0 50 150 251 cases per state Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 enforcement actions). Pandemic Darlings analysis. District codes are DOJ shorthand (e.g. MDFL = Middle District of Florida).

The crime came first. The reckoning is still going.

The money went out in 2020 and 2021, while the programs were open. The indictments and sentencings are still landing in 2026 — and will keep landing. In 2022 Congress pushed the statute of limitations on PPP and EIDL fraud out to ten years, so prosecutors have until the early 2030s to file.

2026-06-19T16:25:25.608925 image/svg+xml Matplotlib v3.10.9, https://matplotlib.org/ 2020 Q1 2020 Q3 2021 Q1 2021 Q3 2022 Q1 2022 Q3 2023 Q1 2023 Q3 2024 Q1 2024 Q3 2025 Q1 2025 Q3 2026 Q1 0 10 20 30 40 50 60 Cases per quarter (DOJ announcement date) relief programs operating Most fraud happened in 2020–21. Sentencings didn't peak until years later — and continue. The crime came first. The reckoning is still going. charged / indicted pleaded guilty sentenced Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 enforcement actions). Pandemic Darlings analysis. Dated by press-release announcement. Congress extended the PPP/EIDL fraud statute of limitations to 10 years in Aug 2022.

Which programs got hit

PPP and EIDL are what people mean by pandemic fraud, and they dominate the case count. The Employee Retention Credit barely registers by number of cases — and quietly owns the top of the dollar leaderboard.

2026-09-30T00:00:00 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ PPP (Paycheck Protection) EIDL (Economic Injury loans) Unemployment insurance ERC (Employee Retention Credit) Restaurant Revitalization 1,779 880 390 30 22 Cases naming each program. A single case can name several — and 710 do. PPP and EIDL are the face of pandemic fraud by case count. But ERC — just 30 cases here — tops the dollar leaderboard. Small in number, enormous per scheme. Which programs got hit Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 enforcement actions). Pandemic Darlings analysis.

Plead, don't fight

When the government already holds your loan application, your signature, and your bank records, the play is to plead. Guilty pleas beat trial convictions more than four to one.

2026-09-24 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ Charged / indicted Pleaded guilty Convicted at trial Sentenced Civil settlement 699 474 108 791 101 Status of each case when DOJ announced it. Almost nobody risks a jury. Guilty pleas outnumber trial convictions 4.4 to 1. When the government already holds your loan application, there isn't much to argue. Plead, don't fight Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 enforcement actions). Pandemic Darlings analysis.

Everything is wire fraud

Wire fraud is the workhorse charge — it appears in well over half of all cases. It's the statute for “you lied to get money through a computer,” which is most of what happened here. Money laundering, bank fraud, and aggravated identity theft fill out the stack defendants usually face all at once.

2026-10-09T00:00:00 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ Mail Fraud False Claims Identity Theft Bank Fraud False Statements Money Laundering Conspiracy Wire Fraud 150 252 394 510 546 617 814 1,540 Charges filed across 2,709 cases. Most defendants face several at once. Everything is wire fraud Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 enforcement actions). Pandemic Darlings analysis.

Most did it alone. The crews were charged with more.

Two-thirds of releases describe someone working alone. The third who teamed up charged out bigger: a $1.8 million median against $1.00 million.

2026-10-09T00:00:00 image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ 0 500 1000 1500 2000 Number of releases Solo actor Conspiracy / co-conspirators 1,842 68% 867 32% share of all 2,709 releases Solo actor Conspiracy $0.0M $0.2M $0.5M $0.8M $1.0M $1.2M $1.5M $1.8M $2.0M Median amount charged $1.00M $1.80M conspiracy cases ~80% higher at the median Most did it alone. The crews were charged with more. Source: U.S. Department of Justice pandemic-fraud press releases, Mar 2020–May 2026 (2,709 releases). Pandemic Darlings analysis. Conspiracy = co-conspirator language in the charge list or release summary; counts cover all 2,709 releases. Medians use the 1,447 distinct schemes carrying a verified dollar figure (solo n=994, conspiracy n=453).

The biggest schemes weren't PPP

The largest schemes the DOJ has charged are tax-credit and health-care-billing operations, several of them amounts merely sought. PPP and EIDL produced the volume; ERC and Medicare produced the whales.

2026-06-19T16:25:26.548929 image/svg+xml Matplotlib v3.10.9, https://matplotlib.org/ $0M $100M $200M $300M $400M $500M TX corporate officers — relief-loan fraud PPP lender-service co-founder Hollywood man — ERC claims TN tax preparer — multi-program scheme Las Vegas woman — ERC refund conspiracy Kabbage (lender) — False Claims Act NJ tax preparer — 1,000+ false ERC returns Lavaca, AR man — COVID health-care fraud AZ man + associates — PPP conspiracy CA respiratory-claims scheme MI test-billing scheme (Medicare/TRICARE) $53M $63M $65M (sought) $80M $100M (sought) $120M (settlement) $124M (sought) $134M $178M $359M $500M ERC tax credit Health-care billing PPP / EIDL multi-program The biggest schemes weren't PPP Largest individual pandemic-fraud cases, by the dollar figure DOJ charged. Color = program. $2.9B NJ trio — false ERC & sick-leave tax credits (amount sought) Nearly 6× larger than any other charged pandemic-fraud case. Source: DOJ press releases; each figure is the Department's own charged/alleged amount. 'Sought' = amount the defendants tried to obtain, not necessarily received. Pandemic Darlings analysis.

The case data carries a “sentence” field, but it captures the statutory maximum a charge could carry, not what defendants got. These two charts use the site's criminal sentences table instead: of its 1,131 records, 784 carry an analysis-grade prison term under the cleaning rule on the sentences chart page.

The median sentence is about three years

Of the 784 prison terms, the median is 36 months, about three years. It rises with the dollar amount in the case, from 18 months under $100,000 to 84 months at $10 million or more, and it differs by fraud family, from 24 months for EIDL borrower fraud to 50 months for identity or unemployment-insurance fraud.

image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ 0 10 20 30 40 50 60 median prison sentence (months) EIDL borrower PPP borrower Platform / lender side Other relief fraud Tax preparer / tax credit Identity / UI 24 mo (n=73) 33 mo (n=396) 41 mo (n=14) 45.5 mo (n=30) 47 mo (n=32) 50 mo (n=239) By fraud family 0 20 40 60 80 100 median prison sentence (months) < $100k $100k–1M $1M–10M $10M + 18 mo (n=103) 28 mo (n=307) 48 mo (n=266) 84 mo (n=63) By dollar amount in the case What pandemic fraud actually costs: the median sentence is about three years Median court-imposed prison term, 784 records. The dollar view uses the 739 with a case amount. Source: Pandemic Darlings criminal sentences table (pandemic-fraud-sentences.csv, release v5): 784 analysis-grade prison terms of 1,131 records, sources dated Nov 2020–Sep 2026; cleaning rule on /charts/sentences-db/.

Stealing more buys more prison — but the line is noisy

Each dot is one sentencing record with a case amount, 739 of the 784. The relationship between dollars and months is real but loose: a fraud under $250,000 can draw more time than one over $5 million, because role and narrative — borrower, tax preparer, identity ring, lender-side operator — matter as much as the number. The 0 records whose amount is a program-wide total above $1 billion are left off.

image/svg+xml Matplotlib v3.9.4, https://matplotlib.org/ $10k $100k $1M $10M $100M dollar amount in the case (log scale; basis varies — loss, proceeds, or amount charged) 0 50 100 150 200 250 300 350 prison sentence imposed (months) median 36 mo Fayne — 210 mo Oudomsine — 36 mo Rufai — 60 mo Named sentences in the table Maurice Fayne — 210 mo on $4.5M Vinath Oudomsine — 36 mo on $85k Abidemi Rufai — 60 mo on $604k Each dot is one sentencing record with a case amount. Small-dollar cases can still draw years; role and narrative matter as much as the number. Stealing more buys more prison — but the line is noisy PPP borrower (n=369) EIDL borrower (n=69) Identity / UI (n=231) Tax preparer / tax credit (n=30) Platform / lender side (n=11) Other relief fraud (n=29) Source: Pandemic Darlings criminal sentences table (pandemic-fraud-sentences.csv, release v5): 784 analysis-grade prison terms of 1,131 records, sources dated Nov 2020–Sep 2026; cleaning rule on /charts/sentences-db/. Dollar basis varies by case (loss / proceeds / charged).

The recovery ledger follows 31 concluded cases through the money stages their documents print, from the amount obtained to restitution, forfeiture and seizure; in 28 a court ordered restitution, and two documents record money returned or collected.

Amount ▾DateState ProgramsStatusCase (links to DOJ release)
Method. Figures come from 2,709 U.S. Department of Justice pandemic-fraud press releases (March 2020–May 2026), parsed into a structured table. Dollar amounts are as charged or alleged. Since data release v9 (2026-09-30) the parsed amount and the program of every release the site's sentencing table covers are that table's, checked against the release, so for those releases the figure is the table's case amount (its amount basis may be loss, proceeds or an amount sought). A release carries a dollar figure here only when that figure is checkable against the release itself: either DOJ put it in the headline (743 releases), or the parsed amount matches a figure in the release's opening paragraph (750 releases). That leaves 1,493 releases — 1,447 distinct schemes once the same prosecution announced by two offices (the same headline and the same amount) is counted once — carrying $7.9 billion. The resolved-case total applies the same rule to the 820 of those releases whose status is pleaded guilty, convicted at trial or sentenced: 799 distinct schemes carrying $4.09 billion, shown as $4.1 billion. The other 1,216 releases stay in the table as cases with no dollar figure: a program total (the $349 billion first PPP tranche, say), a nationwide enforcement total, or an uncorroborated parse tells you nothing about what that defendant was accused of taking. Uncorroborated amounts are excluded, never estimated, so every dollar total on this page is a floor. The press releases' “maximum penalty” language (“faces up to 20 years”) is a statutory ceiling, not a sentence imposed, and is not used for the dollar charts. The sentencing charts instead use the court-imposed prison terms in the site's criminal sentences table (1,131 records), cleaned to 784 analysis-grade terms under the rule stated on the sentences chart page. Where a figure is an amount sought rather than received, it is labeled. The case table below links to each original DOJ release. Sources. Relief disbursed, the ~$2.1 trillion denominator: SBA for the two business programs — PPP (original: sba.gov) (~$800B; $792.6B of approved loans in SBA’s PPP loan-level files) and COVID-EIDL (over $400B of COVID-19 EIDL funds disbursed, per SBA OIG Report 23-09 (original: smallbusiness.house.gov); $370.3B of loan face value alone in USASpending’s COVID EIDL award files) — plus Department of Labor pandemic unemployment payments (~$900B), the base GAO measures its $100–135B fraud estimate against. The components are approximate and rounded (about $800B + $400B + $900B, all money paid out), and the three are added by this page, so the total is best read as a rough scale. Fraud estimates: SBA Office of Inspector General, Report 23-09, COVID-19 Pandemic EIDL and PPP Loan Fraud Landscape (original: sba.gov · stored capture); GAO, GAO-23-106696 (original: gao.gov) (unemployment-insurance fraud, $100–135B) and GAO-23-105331 (COVID-19 relief fraud risk). The fraud estimates on this page are shown by issuer, each with its own date, and are not added together: SBA OIG Report 23-09’s two estimates — its full report (stored capture) prints “more than $136 billion COVID-19 EIDLs and $64 billion in PPP funds” (p. 13) — sit beside the $100–135B unemployment range in GAO-23-106696, held here, which puts the range at about 11 to 15 percent of the unemployment-insurance benefits paid during the pandemic. The EIDL rate is the Inspector General’s own: Report 23-09 prints the $136 billion as “33 percent of total disbursed funds” (p. 16), measured against disbursed dollars (the same report puts COVID-19 EIDL disbursements at over $400 billion). Conviction counts: SBA OIG, Spring 2025 Semiannual Report to Congress (original: oversight.gov) (1,045 convictions since March 2020), and the COVID-19 Fraud Enforcement Task Force, 2024 report (original: justice.gov) (approximately 2,005 defendants who had pleaded guilty or been convicted at trial as of December 31, 2023). Case-level dollars and counts: this page's own table, rebuilt from the DOJ releases it links to.

Pandemic Darlings — primary documents on the pandemic, five years on. Charts and data are reusable; each links back to the underlying DOJ source.

In the archive: Pandemic relief fraud map · How to read pandemic fraud numbers · Fraud numbers are not loss numbers · Court filings · Datasets · All interactives
Back to top