Regression

A study records the same people weekly, then fits an ordinary regression while treating all weekly errors as independent. Why might this be problematic?