> The IRB should not have allowed this to happen.

I see this repeated, but arguments like "You don't get to secretly do things to your subjects." are not sufficient nor is "it's arguably the most important software on the planet". These viewpoints are not agreed upon or codified anywhere that would affect an IRB decision.

"human subjects" qualification -> https://www.fda.gov/about-fda/center-drug-evaluation-and-res... (et al sources)

The notable history of outrage in some communities (did this make the evening news anywhere?) that has been created, may influence future decisions, at best.