It is true and makes sense!
If a certain AV is highly used in the USA, but used very little Europe for example, it's FP rate and detection rate of certain malware families will be different than an AV highly used in Europe vs USA. Some pieces of software may only be available/used in certain regions. Similarly some malware families are only seen in certain regions as well. With the advent of ML and the cloud I would assume that FPs are going to be higher generally speaking, as most AVs seem to be imporing some sort of file reputation system. MD and Smartscreen are a great example of this. Between BAFS and Smartscreen, MD/W10 can be pretty ruthless to new files. To the point that it can be extremely annoying for a developer trying to create a new program.
The problem with FPs in tests (at least the way AV-Comparatives displays them) is you only see the total number as part of the graph/chart. AV-Comparatives use to post an appendix to outline in more detail where the FPs came from, as well as if the file had very low, low, medium, or high prevalence. Using MD as an example, it use to have reasonably high FPs in AV-Comparatives. However, when looking at the appendix, one would see that the bulk of the FPs came from files with very low to low prevalence. Files with medium, or high prevalence were extremely low. Which agrees with what the vendors are telling you. Most people in general use software that is fairly well known to most AV vendors, so the FP rate in the real world for the vast majority of people is going to be really low. However, if you are using an AV that is not widely used in your region and you are either using software that is primarily used in that region, or a file with low prevalence in general, your chances of a FP will be higher.
But..but, all those YouTuber tests always showed me that a large folder of random files will just randomly appear on my desktop one day. They also showed me that it's best practice to just execute each one to see what it is, even though it just randomly appeared....