The question to ask is whether Amazon’s policy and corresponding procedures are reasonably designed and effectively implemented to prevent violations. Will a few things slip through the cracks of a reasonably designed and effectively implemented compliance program? Of course. But a reasonably designed and effectively implemented compliance program will iterate and improve each time something is identified that has “slipped through.” Over time compliance increases while the number of cracks decrease.

Is Amazon’s approach to this policy reasonable and effective given the resources at the company’s disposal? Amazon is a tech behemoth. Surely they could spare a developer or two to come up with a more effective and automated way to monitor for banned items. A process that looks for keywords wouldn’t be hard, especially given how Amazon’s retail system works. Or maybe a system that identifies product listing photos that are similar shape/size to banned items. Amazon has some experience with image recognition and machine learning that would make that much easier to implement compared to most online retailers that allow third-party listings.

Amazon is more than capable of designing and implementing procedures to achieve much more robust compliance with their policy without expending considerably more effort. The only reason they don’t is because they don’t want to - because it hurts their bottom line and (for now) failing to reasonably and effectively implement their policy costs them only the occasional PR ding.

As an Amazon customer I'm fine with them dedicating the legal minimum effort to this. Prices going up or the service getting worse over having to dedicate more resources to banned items would be a loss for me.