These policy questions are framed to always have the authoritarian win.
It's because AI detectors don't have 100% accuracy that they are considered bad.
Working AI detectors are bad.
False cheating accusations are collateral damage to those naive enough to participate.