In 2019 for example, the US had 2.7 million SAR filings. (https://www.fincen.gov/reports/sar-stats, all industry types + year 2019 + all states), compared to a touch above 5,000 white collar federal prosecutions (https://www.cbsnews.com/news/white-collar-crime-prosecutions...). It's hard to directly calculate a "false positive rate" from those things, because some criminals may generate multiple SARs and others may simply go unprosecuted, but I think such a vast gulf is pretty inconsistent with an idea that most SARs report crimes.

It's certainly conventional wisdom in financial circles that a SAR is worth about as much as a user report on a Reddit comment.