The Bureau of Labor Statistics revised its job growth data downward by about 911,000 jobs over a recent 12-month period, indicating that many previously reported jobs did not actually exist, though the exact figure of 'nearly 3 million' jobs over three years is not supported by current evidence.
The Bureau of Labor Statistics has reported nearly 3 million jobs over three years that did not actually exist.

PolitiFact | Donald Trump twists timeline of Bureau of Labor Statistics job data revision
Source: politifact.com
Key Evidence
The BLS announced a downward revision of approximately 911,000 jobs for April 2024 through March 2025.
foxbusiness.comfoxbusiness.comLabor Department revises jobs data showing 911,000 fewer positions through March
abcnews.go.comabcnews.go.comUS economy added 911,000 fewer jobs than previously reported in largest-ever...+1 more
This revision is characterized as the largest ever but limited to a one-year period.
nytimes.comnytimes.comJob Growth Revised Down by Nearly a Million, Updated BLS Data Shows
abcnews.go.comabcnews.go.comUS economy added 911,000 fewer jobs than previously reported in largest-ever...
No credible source confirms a figure close to 3 million jobs over three years being falsely reported.
politifact.compolitifact.comPolitiFact
washingtonpost.comwashingtonpost.comU.S. employers added 911,000 fewer jobs than first reported, new BLS data shows
What the Evidence Shows
Multiple reputable sources confirm that the Bureau of Labor Statistics (BLS) issued a significant downward revision to U.S. Job growth data, specifically reporting approximately 911,000 fewer jobs added between April 2024 and March 2025 than initially reported. This revision is described as the largest ever recorded and reflects adjustments made after more complete data became available.
However, none of the sources provide evidence supporting a claim that nearly 3 million jobs over three years were fabricated or did not exist. The revisions pertain to a one-year period and total just under one million jobs, not three million over multiple years. The discrepancy appears to stem from conflating this recent large revision with cumulative revisions over a longer timeframe or misinterpreting the scale of the adjustment.
Additionally, sources explain that such revisions are normal in labor statistics due to the nature of preliminary estimates being updated with more comprehensive data. While political controversy surrounds these revisions, including criticism and administrative actions, the factual basis is that about 911,000 jobs were overreported in the latest annual benchmark revision, not nearly 3 million over three years.