buffersandbeer

Airport security staff consent to this type of testing at hiring time so testing can be random, and not just anyone can try to sneak a weapon through security to see if it's caught as "a test".

Perhaps a similar approach that allows randomness with some sort of agreement with the maintainers could have prevented this issue while preserving the integrity of the study.

Has an issue been opened in their github for this? I looked quickly and didn't see one, but would love to track it.

Probably due to a training bias