Generative artificial intelligence (GenAI) has had an impact on university education and also to the software industry. This can be seen in many different courses where GenAIs can complete exercises, exams, and even whole courses by themselves. This study aims to find out how effectively different GenAI tools can complete exercises in programming courses. Our objective is to find how GenAI’s performance varies across courses and what differences there are in results between multiple GenAI tools. During the study, five programming courses were evaluated using five different GenAI tools. These tools were able to solve 43–90% of programming assignments and 80–100% of quizzes. In practice, the result is that any of the selected GenAIs could have been used to pass the majority of all the tested course parts. Results indicate that we have to rethink how courses’ tasks are designed and that courses need to have an element where learning is verified in a controlled environment to verify that we are educating software professionals for the industry.