Some more grandmas dying would be "acceptable" damage. However, that isn't the problem.

The problem is this: The risks of caution.

1-line summary: if the good guys delay their projects to make them safer, the bad guys are more likely to win.

The video's "abstract":

It is commonly thought that caution in the initial development of machine intelligence is associated with better outcomes - and that things like extensive testing, sandoxes, and provable correctness are things that will help to produce safe and beneficial synthetic intelligent agents.

In this video, I cast doubt on that idea, by exhibiting a model in which delays caused by caution can lead to much poorer outcomes.