I don't think the problem here is with the QA workers. This is clearly a bug in the underlying internal test framework, which the QA workers have very little info about. And the likelihood of a given worker seeing enough of these errors for them to realize it's an issue with the system and not the app is fairly low.
The real issue is that the engineering team who maintains the internal app checking system 1. needs to have infrastructure to detect abnormal amount of a given error and 2. need to notify the QA team so the QA team can communicate it with the devs, rather than just blaming the apps.
You'd think people would look at the initial healthcare.gov mess and make some conclusions.
One of which should maybe be "Don't strictly isolate teams, with unowned space between their output and the next team's input, and no method by which post-delivery failure reflects back on them."
The real issue is that the engineering team who maintains the internal app checking system 1. needs to have infrastructure to detect abnormal amount of a given error and 2. need to notify the QA team so the QA team can communicate it with the devs, rather than just blaming the apps.