How to treat AI systems that pretend to be humans?

The question makes sense if you consider to have an AI assistant that is pretending to be human while the program running in the background is getting more and more sophisticated. On the other hand, with all the flaws in most modern AI systems that pretend to be humans (the popular example being chatgpt) seems to be too far from the real AI systems like HAL.<br><br>Personally, I will treat them as objects that I pay for the services and not more. Let's say a coffeemaker or a laptop. It is in the companies' interests to make us humanize them but let's don't fall into this and try to stay objective to keep it safe for the mankind.