Should we hold an autonomous robot that has caused harm to others criminally liable? The emerging technology of autonomous robots touches upon our perception of the foundations of criminal liability in an unprecedented way. Essentially, it raises the question whether we have to redefine personal autonomy as a foundational requirement of criminal liability in order to tackle digital developments. This paper analyses the application of conventional doctrines of criminal liability to the scenario of an autonomous robot harming others. Based on the analysis, the paper proposes a partial shift of risks associated with the use of robots from the individual user onto society by limiting individual criminal liability in order to reconcile the competing interests. The paper further shows why holding robots criminally liable is, at least as things stand today, not a feasible solution. Instead, the paper suggests alternative legal measures that might be better suited to address the challenges and close a potentially emerging liability gap than the (exclusive) reference to criminal law.