Nobody’s Checking the Box: What DOL’s Own H-1B Disclosure Data Reveals About Its Quality Control

Nobody’s Checking the Box: What DOL’s Own H-1B Disclosure Data Reveals About Its Quality Control

We ran two basic data-integrity checks against the Department of Labor’s H-1B LCA disclosure archive — the same public dataset guestworkervisas.com’s search tools are built on — and found a paper trail with almost no gatekeeping. Employer identity isn’t standardized (Amazon’s own visa-processing subsidiary filed under nine different spellings of its own name), and the one checkbox meant to flag employers with a history of willful labor-law violations is filled in so inconsistently that none of the employers DOL has actually barred from the program ever checked it. Data: guestworkervisas.com’s local mirror of DOL/OFLC H-1B LCA disclosure data, 2008–2026 (3,831,829 filings), cross-referenced against DOL’s public H-1B debarred-employer list as of August 2026.

Duplicate employer-name entries
24,703
13.1% of all 188,152 distinct employer-name strings in the archive are the same company typed differently — not different companies
Real employers fragmented across spellings
20,604
Split across 2 or more “distinct” name entries by nothing more than a comma, a period, or an extra space
Filings marked “willful violator”
0.03%
1,142 of 3,831,829 LCA filings, scattered across 671 employer names with no pattern matching real enforcement
Actually-debarred employers who ever checked that box themselves
0 of 4
Every employer DOL has currently barred from the program marked “No” or left it blank on all 81 of their combined filings
Employer identity has no canonical ID — not even for Amazon

DOL’s LCA disclosure files identify employers by a free-text name field with no employer ID, no deduplication, and no standardization pass. We normalized every employer name in the archive as conservatively as possible — uppercase, strip commas and periods, collapse whitespace, nothing fancier — and 24,703 of 188,152 “distinct” employer names collapsed into a name already used by another row. That’s a floor, not a ceiling: this check doesn’t touch abbreviations, “US” vs “U.S.A.”, or legal-suffix variants like “LLC” vs “L.L.C.”, all of which would fragment real companies further.

Amazon Development Center US, Inc. 9 spellings US Electronics, Inc. 8 spellings Dish Network LLC 7 spellings Everest Consulting Group, Inc. 7 spellings ICS Global Soft, Inc. 7 spellings Intellectt, Inc. 7 spellings Morgan Stanley & Co. LLC 7 spellings Penske Truck Leasing Co., L.P. 7 spellings Teradata US, Inc. 7 spellings Amadeus North America, Inc. 6 spellings
Most-fragmented name in the archive
Other employers, ranked by spelling-variant count
Also in the archive with 6 spelling variants apiece, purely from punctuation and spacing: Charter Communications, Inc., GlobalFoundries US, Inc., Michael Kors USA, Inc., and SmarTek21 LLC — the last of which shows up as literally the same string with one to five trailing spaces appended. These aren’t small or unfamiliar companies; they’re firms with dedicated immigration counsel filing thousands of LCAs, and even they can’t get their own legal name typed the same way twice in a federal disclosure system.
The one checkbox meant to catch bad actors is functionally noise

Every LCA asks the employer to attest whether it has previously been found a “willful violator” under the H-1B program — a designation that, once made, is supposed to be disclosed on every filing for five years. In the full archive, that field holds six different raw values for what should be a single yes/no answer (Yes, Y, No, N, N/A, and blank), and only 1,142 of 3,831,829 filings (0.03%) are marked positive at all. Tata Consultancy Services’ own filing history shows why that number is meaningless as a signal rather than reassuring:

Fiscal year Total TCS filings Marked “willful violator” Share
FY2020 11,998 5 0.04%
FY2021 5,259 0 0.00%
FY2022 11,172 10 0.09%
A genuine five-year disclosure obligation would show up as “Yes” on every filing in the window, not 5 out of 11,998 in one year and 0 out of 5,259 the next. The same pattern holds for other large filers who show up as “positive”: Amazon.com Services LLC has exactly 1 filing marked “Yes” out of tens of thousands, and Meta Platforms has 2, both prepared by the same outside law firm a year apart. These read as isolated data-entry slips by individual preparers, not disclosures of real enforcement findings — which cuts both ways: the field is just as unreliable when it says “no” as when it says “yes.”
None of DOL’s actual debarred employers ever flagged themselves

DOL’s public H-1B debarred/disqualified employer list currently names four companies as confirmed willful violators. We checked each one’s own historical LCA filings for the self-attestation the form is supposed to require of them.

Employer Current debarment period Total filings on record Self-attested “willful violator”?
GowraTech, LLC 5/12/2025 – 5/11/2027 10 No / N, every filing
Renotek Group LLC (also filed as Renotek Solutions LLC) 8/8/2025 – 8/7/2027 67 No / N, every filing
Seeloz, Inc. 3/4/2026 – 3/3/2028 3 No / N, every filing
Sherwood at Mount Dora, Inc. (dba Sherwood Academy) 5/26/2026 – 5/25/2028 1 N/A
This isn’t a contradiction — debarment is a Wage and Hour Division enforcement action that happens after the fact, so an employer’s older LCAs predating the finding wouldn’t be expected to show it. But it means the self-attestation field provides zero advance warning in the one place it would matter most, while simultaneously flagging unrelated Fortune 500 filers through what look like ordinary clerical mistakes. As a public signal, it fails in both directions at once.
What this means

This is the same dataset every H-1B search tool on this site — and most of the H-1B journalism published anywhere — is built on. It has no employer ID, no name standardization, and at least one self-attestation field that’s essentially decorative. The practical consequence is that any “who are the top H-1B sponsors” ranking, including ones built from this exact archive, is undercounting real concentration by splitting single companies across multiple name entries, and any claim that leans on the willful-violator field as evidence — in either direction — is standing on a field nobody appears to be checking. None of this requires a conspiracy; it’s what happens when a federal disclosure system accepts free-text employer names from thousands of different law firms and staffing companies with no validation layer in between. The debarment list itself, by contrast, is small, curated, and appears reliable precisely because it’s DOL’s own enforcement output rather than employer self-report.

Data: guestworkervisas.com’s local mirror of DOL/OFLC H-1B LCA disclosure data, 2008–2026 (3,831,829 filings, 188,152 distinct employer-name strings) · DOL Wage and Hour Division H-1B Debarred/Disqualified List of Employers, effective August 1, 2026.