The harm from AI is already here. What can the US do to protect us?