Why would real accuracy be difficult to determine? Clinicians get to test these on their patient populations and validate against other high accuracy methods (point-in-time serum draws). For instance, Dr. Peter Attia says he gets a useful fraction of his clinic to use CGMs (as well as personal use), and over years sees a strong signal.

There are many other Drs that have gone on record.

Anecdotally - I helped a T1 diabetic friend set up Loop using a Dexcom G5 and then a G6. Over the course of many months of use, the G6 has proved MUCH more reliable and accurate than the G5, which resulted in far better Looping. The Loop community as a group seem to confirm this increased accuracy.