It does, and I think the problem is more that Amazon's internal processes and tribal knowledge are a bit clueless when it comes to fraud. Their perspective, historically, has been to allow as many products as possible on the site, because one of the way they get and retain customers is by 'having everything'. Their metrics don't incentivize large purges. I think it's a bit of an arms race between fraudulent sellers and the retail org's processes and engineering, and unfortunately the latter iterates much slower so they can't keep up.
Also, the warehouse processes at least are heavily designed around catching mistakes like mislabeling, but less so intentional fraud like fakes (or - they were a few years ago, imo). And they tend to treat fraud like a big data problem ('what can we do to lower this statistic') instead of case-by-case: 'we need to get rid of every fraudulent seller'. Which, like, if you try really hard to move a needle from, say, 80% to 95%, that ultimately means you are vocally okay with 5%, which is stupid.
Also, the warehouse processes at least are heavily designed around catching mistakes like mislabeling, but less so intentional fraud like fakes (or - they were a few years ago, imo). And they tend to treat fraud like a big data problem ('what can we do to lower this statistic') instead of case-by-case: 'we need to get rid of every fraudulent seller'. Which, like, if you try really hard to move a needle from, say, 80% to 95%, that ultimately means you are vocally okay with 5%, which is stupid.