If you hold everything else constant, an increase in Type I error means a decrease in Type II error. This is known as the error trade-off in statistical hypothesis testing.