Is tricking an AI system hacking under the Cybercrime Convention?

In a 2018 paper, Calo et al. discuss whether “tricking a robot” constitutes hacking. They focus on five scenarios under the U.S. Computer Fraud and Abuse Act (CFAA). I thought it would be worthwhile to look at those same scenarios from the perspective of the Council of Europe Cybercrime Convention (also known as the Budapest Convention). (I refer to AI systems rather than robots, but we mean the same.)

I conclude that four of the five scenarios should be criminalised under the Convention, but one should not.