The Department of Government Efficiency (DOGE) has finalized an extensive audit of the Social Security Administration’s (SSA) death records, resulting in the classification of around 12.3 million individuals as deceased. This review, described as a long-overdue administrative correction, follows years of accumulating discrepancies within one of the country’s most critical identity databases.
According to Newsweek, many of these records listed implausibly high ages—some exceeding 160 years—raising concerns over the integrity of public data. While the cleanup operation impacts how records are maintained, it reveals little about benefit fraud or misuse, shedding limited light on the broader implications for Social Security.
Historical Discrepancies in Death Records
DOGE launched the initiative following the discovery of extraordinary inconsistencies in the Numident database, which has tracked Social Security numbers since 1936. According to data shared by DOGE on X (formerly Twitter), many records listed individuals as being far older than any verified human lifespan.
The cleanup effort revealed:
- 7.2 million individuals listed between 120 and 139 years old
- 4.8 million between 140 and 159
- 124,000 over 160 years old
- And at least one case showing an age between 360 and 369 years
The entire operation spanned 11 weeks, with DOGE stating on X:
After 11 weeks, @SocialSecurity has finished this major cleanup initiative: ~12.3M individuals aged 120+ have now been marked as deceased.
This project builds on a 2015 SSA review, which had already identified 6.5 million individuals over 112 years old whose death records had not been updated. SSA policy suspends automatic benefit payments to anyone listed as being 115 years or older, regardless of reported death status.
Musk Spotlight Prompts Federal Review
Public awareness of the data issue surged in February, when Elon Musk, head of DOGE, posted a spreadsheet on X showing individuals aged 100 and above still marked as “alive.” The sheet flagged those entries with the tag “FALSE!” in the death status field, suggesting the SSA system had not registered their passing.
Musk also noted that over 10 million entries in Numident belonged to individuals reportedly over 100 years old, reinforcing the urgency of the issue. While these records did not confirm benefit fraud, they pointed to major gaps in data reporting and record hygiene.
Experts Highlight Structural Causes
According to the Brookings Institution, these anomalies stem from legacy practices in federal documentation. In a public report, the think tank stated:
The absence of a death date doesn’t necessarily mean the person is still alive; it may simply mean the death was never reported.
They added that:
Computerized death records were not introduced until 1962, meaning information on deaths prior to that year is often incomplete or inaccurate.
This highlights the impact of manual, paper-based systems used in early Social Security administration and the challenges of ensuring consistent updates across decades of data.
No Evidence of Fraudulent Benefits, Says SSA
Following Musk’s public disclosures, the SSA issued a statement on March 5 to clarify:
The data reported in the media represent people who do not have a date of death associated with their record. While these people may not be receiving benefits, it is important for the agency to maintain accurate and complete records.
The agency emphasized that these entries reflect administrative oversights, not direct evidence of fraudulent payments.
Unresolved Anomalies Remain
Despite the scale of the cleanup, DOGE acknowledged that several complex cases remain unresolved. These include instances where individuals have multiple conflicting birthdates in their records. DOGE confirmed:
Some complex cases remain, such as individuals with 2+ different birth dates on file. These will be investigated in a follow-up effort.
This continuing effort reflects the bureaucratic challenges inherent in modernizing longstanding government databases, especially when they span both analog and digital eras.
