A $5.7 billion startup wants AI to cut government benefit fraud. Experts aren't sold yet. (www.businessinsider.com)

Checkr, a San Francisco-based startup valued at $5.7 billion, is setting its sights on securing government contracts to automate identity verification for welfare programs like Medicare and Social Security. CEO Daniel Yanisse highlighted the intention to combat fraud and ensure that the right individuals receive benefits. However, experts in AI and safety raise concerns about the viability and safety of automating such critical government functions, citing the legal and technical challenges involved. Current estimates indicate significant improper payments within Medicare, amounting to nearly $29 billion in 2025, underscoring the need for improved verification systems. While Checkr has a successful track record with identity verification in private sectors, experts warn that previous government automation attempts have sometimes led to disastrous outcomes. Issues highlighted include a failed welfare project with IBM in Indiana and the controversial Robodebt scheme in Australia, both of which resulted in significant harm to the affected populations. Experts like UC Berkeley's Stuart Russell and Emory University's Ifeoma Ajunwa advocate for cautious approaches with strong oversight, emphasizing the need for evaluation and accountability when deploying AI technologies in government, particularly in sensitive areas impacting people’s lives.