AI is growing to play an increasingly important role in the decision environments of human agents. One of the key challenges here is AI influence - an aspect that seems largely unaddressed in current debates around AI Ethics. In this talk, I aim to shed some light on the notion of AI influence and the problems that surround it. In the first part of this talk, I will outline what exactly I mean by AI influence. For this, I will introduce what I call the objectivity-fallacy, and will differentiate between cases of intended and unintended AI influence. The second part of this talk will then focus on the ethical implications of unintended AI influence. I hope to emphasise an important change in the power dynamics in human-AI interaction, and will outline how unintended AI influence changes the roles and thereto related responsibilities in human-AI interaction.