The Department of Government Efficiency (DOGE) has launched a high-speed effort to rewrite the aging codebase behind the U.S. Social Security Administration — a project involving over 60 million lines of legacy COBOL code. Experts warn the move could pose serious risks to the payment systems that support more than 65 million Americans.
The Social Security system relies on decades-old code and a patchwork of interdependent databases. While modernization is long overdue, DOGE’s plan to rebuild the entire architecture in just a few months has alarmed software engineers and policy analysts. “This isn’t just a website redesign,” one government technology expert told Ars Technica. “It’s a payments infrastructure with life-altering consequences if something goes wrong.”
Critics point to the potential for data integrity issues and system outages during the transition. The COBOL-based environment includes custom workflows tied to income verification, benefits adjustments, and cross-agency data sharing. A rushed rewrite could introduce bugs that result in underpayments, overpayments, or missed checks for millions of recipients — many of whom rely on benefits to survive.
DOGE has suggested the use of generative artificial intelligence to accelerate development, though experts emphasize that any automation must be paired with rigorous testing and phased rollout plans. Past modernization efforts by the SSA were deliberately slowed or paused due to COVID-19 and the high risk of system disruption.
While modernizing Social Security is critical, analysts warn that compressing a years-long infrastructure overhaul into a matter of months could create more problems than it solves — especially when those problems affect the nation’s largest public payment system.
