I have seen several people raise concerns about integrity of Segment leaderboards, mostly due to the wrong Activity type (using an E-bike on a "ride" or a bike on a "run") or leaving Strava running at the end of an Activity and hence recording a subsequent car journey.
For what it's worth, here's my approach to trying to help maintain Segment integrity in my area. In the last 5 years I have attempted about 6200 bike Segments, all within a few miles of where I live. I use the excellent VeloViewer companion for Strava which allows me to get lists of all my Activities and Segment efforts. It allows me to export data to a spreadsheet and so I can have a spreadsheet of all the Segments I have attempted, complete with my leaderboard position, best time, KOM time, etc. Generally the Segment data is very good with very few suspect KOMs. Roughly once per week I download the data to a spreadsheet and do a comparison with equivalent data from the week before, in particular looking for changes in KOM times. This week, for example, it showed about 30 Segments had improved KOM times compared to last week. About half of these looked genuine performances (small improvements, high heart rates, and credible max speed or power). The other half were from 4 Activities that all looked suspicious - such as multiple KOMs with large improvements, no heart rate data and incredible sustained high speeds (e.g. 55 mph uphill). Hence by flagging these Activities I have hopefully made a contribution to maintaining the integrity of the Segments in my area.
