You should always check your absorber solution just before it is used by running the titration on it. This is called running a blank. It should have an m alkalinity that is small (preferably zero). Sometimes it is small enough that you can simply subtract the initial m alkalinity from the sample's final m alkalinity to get a more reliable result.
Running a blank is a good idea to consider for all analytical tests in the lab.
As the initial solution's m alkalinity creeps-up to where it is say 1 or 2 percent of the typical m alkalinity for the sample, you should replace it with fresh absorber. The exact level that you use as a maximum allowable initial alkalinity for the starting solution before you deem it unusable depends on the amount of error you are willing to tolerate. (It's rarely zero.)
The good thing about having a lab is that you can test your ideas about why a problem might be occurring and also test to determine how much error you are willing to tolerate, but remember you will always have some error in every test.
On the CO2 formation, it occurs in all of your titrations during the m alkalinity part of the test, whether the m alkalinity is from degradation of the starting solution or from legitimate m alkalinity picked-up in the scrubbing process. (Provided some of it is caused by HCO3, which it, in part, usually is).
When the amount of CO2 is small you won't see the bubbles because it remains in the hydrated form: H2CO3, ie H2O-CO2, unless you overshoot the the m alkalinity end-point significantly.