They even address what you are saying in the article. They mention the looming danger of weaponized AI which is exactly when someone has altered the program to do harm.