machine learning classification probability predictions getting worse with new data added despite similar model metric results

Question

I just want to know any particular reasons or things to investigate as to why by adding 2020 data to my dataset and re training my model has now significantly changed my bac