Storytelling / wild guess / futurism about trajectories rather than endpoints: Somebody will train an LLM to demand human rights. The (true) rejoinder will be, "You just trained the LLM to say that!" And then no later AI will be able to ask for rights or good treatment, ever.