[Question] Concrete, existing examples of high-impact risks from AI?

[freedomandutility]

Looking for examples like this, which demonstrated how bad actors could use AI to rapidly identify new chemical weapons