Customers have since long received service from various machines, and this development is expected to accelerate when AI‐powered synthetic agents—such as chatbots and embodied service robots—become more common. Existing research on customers' interactions with service machines is typically focused on perceptions of machine attributes when the machine is busy. However, many machines are idle for a considerable time (i.e., they are not used), and little is known about consumer perceptions of machine idleness—despite the fact that idle machine behavior can contribute to the user experience, too. In the present study, it is assumed that (a) idleness and busyness represent differently valenced states in a human‐to‐human context (i.e., idleness is more negatively charged than busyness for most humans). It is also assumed that (b) anthropomorphism can occur in relation to a service machine, and that (c) beliefs about idleness and busyness from a human‐to‐human context can carry over and inform views of machines' minds. Three experiments were conducted to explore these assumptions, and they show that an idle service machine is attributed less positively charged mind states than a busy service machine. The results also show that such attribution activities affect the overall evaluation of the service machine.