Brazil Family Blames AI System for 32-Year-Old's Death After 5-Day ICU Delay
Updated
Updated · Gizmodo · Jun 15
Brazil Family Blames AI System for 32-Year-Old's Death After 5-Day ICU Delay
3 articles · Updated · Gizmodo · Jun 15
Summary
Rebeca Cardoso Tenente Molina, 32, died hours after reaching an ICU, after waiting five days for a transfer her family says was delayed by Minas Gerais' AI-backed bed allocation system.
Core-MG allegedly scored Molina's condition at 6.8 instead of 10 even as her test results worsened, her twin sister told MG1, effectively overriding doctors who viewed her case as critical.
Molina was hospitalized June 2 in São João Nepomuceno with suspected gallstones and was eventually sent 300 kilometers away for intensive care; her cause of death is listed as septic shock, with botulism still under review.
Minas Gerais' Health Department said Core-MG did not change bed-allocation criteria and that placement depends on clinical need and bed availability, not just proximity.
The case adds to wider scrutiny of medical AI, as research has found some systems can entrench healthcare bias or underestimate emergencies despite promising diagnostic results in other settings.
If a hospital's AI makes a fatal mistake, who is legally responsible?
When a doctor and an AI disagree on a life-or-death decision, who should win?
A Fatal Delay: How an AI Severity Score of 6.8 Sparked a Crisis in Brazilian Healthcare
Overview
Rebeca Cardoso Tenente Molina died after a five-day delay in her transfer to an Intensive Care Unit, with septic shock listed as the cause of death. Her family claims that the state-run Core-MG AI system played a direct role by assigning her a severity score of 6.8 instead of the 10 they believe she needed. This lower score allegedly caused the critical delay in her ICU transfer. The case highlights concerns about how AI systems like Core-MG influence life-or-death decisions in healthcare, especially when their assessments may not reflect a patient's true condition.