To start, looking into the source behind this claim reveals it was made by an individual account, without any reputation as a credible journalist. The account also seems to be very active in political discourse, commonly voicing negative views on Democracy as is, as well as American political and economic practices. While those are fair and justifiable views, they introduce a possible bias to the equation, as this person will likely be a critic of privatized healthcare. As far as the actual claim, it seems to stem from a lawsuit filed in Minnesota in 2023. The lawsuit was filed by the family estate's of two elderly patients denied life-saving care by United Healthcare, and grew to become a class action lawsuit. It is yet to be settled, so nothing can be said concretely, but multiple articles, linked below, back up and fortify aspects of this claim posted on the platform X. They outline accounts from employees, studies, and other evidence that shows that an AI model in use since 2019 by United Healthcare has a roughly 90% error rate, and was used to override primary physicians to deny care provided by patients' care plans. So while this is not a settled case, and the user making the claim is predisposed to be biased against private healthcare, evidence outlined in the below linked articles point to the first part of this claim being true. As far as the claim that this AI model resulted in the death of millions, as well as the claim that CEO Bryan Thompson was a serial killer, much less can be said. It is difficult to know exactly how many patients this effected, and attributing these deaths as murders by the CEO is certainly a gray area. On the whole, the claim appears to be true, but interjections of opinion make the statement more radical and less objective.
https://www.cbsnews.com/news/unitedhealth-lawsuit-ai-deny-claims-medicare-advantage-health-insurance-denials/
https://arstechnica.com/health/2023/11/ai-with-90-error-rate-forces-elderly-out-of-rehab-nursing-homes-suit-claims/
