An investigation by Africa Uncensored, Lighthouse Reports, and The Guardian has found that Kenya's new national health insurance system — launched in October 2024 as a flagship reform of President William Ruto — is overcharging its poorest citizens through a flawed algorithmic assessment tool. The system, known as the Social Health Authority (SHA), uses a machine-learning algorithm based on proxy means testing, estimating household incomes from indirect indicators such as roofing materials, toilet type, and livestock ownership, then setting annual premiums accordingly. Auditing the system against thousands of real households revealed it consistently overestimated the incomes of poor families while underestimating those of wealthier ones — a trade-off that, according to a former government health economist, was a deliberate design choice. The consequences have been severe: many low-income Kenyans now face premiums amounting to between 10% and 20% of their earnings, and those who cannot pay risk being turned away from public health facilities. Of the roughly 20 million people registered under the SHA, only around 5 million regularly pay their premiums, underscoring the scale of the crisis. Critics note that similar proxy means-testing systems, often promoted by the World Bank, have repeatedly failed across Africa, Asia, and Latin America, with error rates in some countries exceeding 80%.