How useful can machine-learning be in dealing with vectors of disinformation such as deep fakes or bots, and what are the implications of AI-powered fact-checking and deprioritising systems for media pluralism and freedom of expression?