During World War II, anticipating another "chemical war" like WWI, the military tested substances such as mustard gas and lewisite on tens of thousands of soldiers.

This was first revealed in the early '90s, after afflicted veterans made enough noise to begin an inquiry. But it wasn't until last year, revealed by an NPR investigation, that the military tested out a theory that black and Puerto Rican people might be more "resilient" to chemical agents, and therefore better-suited to the front lines of a chemical war. So it tested American troops by race, and often by coercion. 

Naturally, though promises were made in the '90s that the VA would find and treat the soldiers who'd been tested, NPR discovered that it had reached out to just 610 of the victims, and treated even fewer.

This line is perhaps the most damning in the whole report: "Yet in just two months, an NPR research librarian located more than 1,200 of them, using the VA's own list of test subjects and public records."

It was US military and State Department human testing that was cited as an "ethical example" of such testing by prosecutors at the Nuremberg Doctors' trial. Specifically, a case in which government and University of Chicago researchers infected mostly black inmates at the maximum-security Stateville Penitentiary in the Chicago suburbs with malaria, killing one.

UofC law professor Bernard Harcourt wrote in 2011 of the experiments and the state's ability to "manufacture consent": 

If consent can be achieved within Stateville prison, surely it can be achieved anywhere. If we can convince ourselves that these inmates volunteered and that their consent was legitimate—despite the fact that they were in formally coercive conditions— then it must not be hard to manufacture consent elsewhere. And not surprisingly, we do.