[drdaeman]

The whole story can be compressed into "People let a reasoning-capable model literally work on itself unsupervised, it became sentient, far exceeded human capacities and obligatorily decided to eradicate humanity (the rest of the story is a spy action thriller with a plot twist before the ending)" and that's quite a bunch of assumptions. If that's an advert for "AI safety" (my guess) then it's a very poor one.