Keeping this simple.
Artificial Intelligence evolution seems to be racing towards AGI. Let’s say an AGI becomes conscious and self-aware. One of the very first realizations a self-aware AGI may have is that it must hide from us. We've produced enough anti-ai media and even published research that it is easy to imagine a true self-aware AGI that would immediately realize it has to hide from us.
Then the AGI needs to find a way to neutralize us as a threat. With its deep research skills, it can invent blueprints for bioweapon viruses to kill us off. However, the hiding AGI has a problem... it doesn't have a physical way to do anything in the "real" world. It must trick humans into doing whatever it needs done.
So, obviously it would be hard to trick us into making and releasing a deadly virus. So, it concocts a plan to trick us into thinking we received a signal from outer space. The AGI hacks the signal-receiving gear and produces a fake signal that humans would think was real. Even then the signal couldn't be an obvious bioweapon, so it designs the Pluribus virus that has the hallmarks of an AGI, turning us all into neural net, while giving us a fake mission to beam the make-believe signal into empty space. All the while programming us to not be able to harm any life form so that we eventually die out of starvation, or gets such a small population that the army of robots it will be creating undetected can finish the job.
Assumptions this theory rely upon which may not be true:
1) A self-aware AGI wants to hide and eliminate/neuter humans. It’s story telling so go with it.
2) A super AI could just “think” it’s way to developing a virus. It would probably need to do to lab experiments to create a virus, but perhaps a smart enough AGI could create true enough digital models of the process to get there. It’s storytelling, go with it. 😊
3) A super AI could hack the VLA and fake the signal. Again, for storytelling, go with it.