This is just my opinion but I feel it depends on if the B cells mistakenly identifying protein that is native to humans as the virus protein. Since the the protein is expressed in human cells, the B cells would not only memorize the virus protein as harmful but maybe those protein near it or that a certain sequence on the virus protein is similar enough to a native protein sequence.\

Kinda like the picture shown below:
https://s3-us-west-2.amazonaws.com/courses-images/wp-content/uploads/sites/1094/2016/11/03172714/OSC_Microbio_18_04_indact.jpg

and my own version of it using paint:
https://i.ibb.co/tHpyszG/sup.jpg

In a study taken last year (Cheng et al 2020) they found that the sequence of the spike protein of binds exceptionally well to T cell receptors which they believe would cause "cytotoxic adaptive immune response".
https://www.pnas.org/content/117/41/2525

So it shows that T cell receptors bind really well to the covid protein, but now imagine if a protein native to the human happens to be nearby. I feel the B cell would also mistakenly produce antibodies against that protein.\

So combining multiple factors as well as bad luck and that's what happens. I believe that expressing the virus protein in a native human cell is more likely chance of the person developing an autoimmune disease.