AI assistants have entered various spheres of activity including education, work life and personal life. The extensive linguistic capacities of these systems lead many to treat them as helpful partners enabling them to find out more, perform intellectual tasks and at time to explore their emotions. But should we treat them as human-like agents or human usable tools? The answer depends on whether we should include them in the linguistic community as fully fledged linguistic agents. Whether we should or not depends on the correct characterization of what they do, on whether they use and understand a language. I will consider what it would take for them to do so.