It will be too easy to create an army of misinformation channels. To make it even more dangerous, if you add elements of machine learning (ML) to the production of these channels, you are providing a mechanism for continuous improvement. When the goal is to misinform people, this is a bad thing. The benefits are that potentially, at least initially, this type of content will have signatures that will be able to be picked up for easy filtering. Maybe even be automatically flagged as automated or computer-generated content. It is a bit like the advent of SPAM, it used to be costly to send mail to everyone, then with email, anyone could send out a million messages with little cost. Right now, it is costly to create unique content especially with human delivery, AI news will have the potential to become the SPAM of news channel communications.