It's really hard, if not impossible for a thermocouple to read wrong. It's based on a very simple joint between two metals, and basically if they are touching at all, it works.
If it's open or a dead short, you'll get wrong values, but they won't change with heat. I assume yours do vary?
The failure we have seen is the tip of the probe being eroded, exposing the actual thermocopule junction inside. This causes higher readings in many cases. The failure should be obvious with the probe removed.
If the two probes reading high look like they are in OK physical shape, check all your wiring. Also, are you verifying that it doesn't follow a specific channel on the EMS, not a specific thermocouple?
