A team of academic researchers has demonstrated that it’s possible to possible to closely mimic legitimate voice commands in order to carry out nefarious actions on these home assistants.
Source: Threadpost