The reality may be that the superalignment team, in practice, had nothing to do. The real AI risks are social impacts like deep fakes and job losses, not hypothetical dangers of AI super intelligent beings getting all HAL 1000 on us. OpenAI has achieved amazing things, but perhaps they are seeing that the AGI they imagined is not around the corner, and devoting high powered resources to trying to contain it is a waste of money.