Assuming the students are making the claims honestly, this indicates to me that he SHOULDN'T be teaching, in the first place.

Yes, the university has a responsibility to educate its faculty on ChatGPT and the warning signs for which educators should be on the lookout...but they need to make it clear *not* to accept bland ChatGPT tests as gospel when it comes to determining a student's potential dishonesty.

If this professor was willing to fail half of his students based solely on a simple ChatGPT analysis, then it's clear he ISN'T operating in good faith when it comes to trying to weed out the cheaters and plagiarists from the students who synthesize content honestly.