There are two types of AI safety:

1. The companies that make AI only care if AGI can happen. Terminators can nearly wipe out humanity, but that is A-Ok with them. They don't care about current or near current future people. They are far future. They specifically have no empathy for anyone except whatever exists in a million years.

2. Everyone else who thinks that AI shouldn't be used to kill anyone

The people in group 1 love that people mistake what they say for meaning AI safety as group 2 sees it.

Just more effective altruism bullshit in another shit bag

Safety is pretty easy: universal corporate death for any lab found negligent.