Having worked on Microsoft Band, I can tell you that accuracy is a problem with UV exposure for a number of reasons, none of which involve the sensor itself.
Mostly it is around how people wear it, and was the sensor always facing the sun, people going in and out of the shade, long sleeves, and so forth. We settled on # of minutes of high exposure, I forget what index we considered "high" but basically it if it was above a certain threshold we considered you "in the sun".
The goal was sunscreen application reminders, no way would we have trusted it to tell people they had "enough" sun!
But in fairness, the sensor doesn't work with how people use it. We could make a sensor that straps to user foreheads and requires them to always face the sun. When integration fails, we can point the finger either way, or just say it wasn't a good fit.
And if there were consumer devices, the next problem would be accuracy. Based on a bunch of reading I did several years ago (so maybe out of date), consumer excercise trackers were very innaccurate in many ways. Exercise misinformation devices.