Not inside a body, no. These aren't the most accurate tests (the big failure mode has nothing to do with the science, it's that the little clamp falls off the finger), but they are very mature technology with well characterized behavior.
Basically everyone who's been under anaesthesia of any kind (even stuff like wisdom teeth removal) for the past 3-4 decades has had one of these on them. We know how they work.
I read that despite the simple measurement mechanism ("shine a light through a finger"), the accuracy is generally pretty high (+/- 2%). This document from the American Thoracic Society does state some corner cases though [1]. Quote:
"How accurate is the pulse oximeter?
The oxygen level from a pulse oximeter is reasonably accurate. Most oximeters give a reading 2% over or 2% under what your saturation would be if obtained by an arterial blood gas. For example, if your oxygen saturation reads 92% on the pulse oximeter, it may be actually anywhere from 90 to 94%. The oximeter reading may be less accurate if a person is wearing nail polish, artificial nails, has cold hands, or has poor circulation."
One very important thing to note is that the Oxygen - Hemoglobin dissociation curve has a sigmoidal shape. It is not linear. After a certain point, around 90 %, there is a very sharp drop-off.