I would be a lot more worried about an AI that does that on purpose.

What happens when the first ransomware is deployed by AI, on behalf of a user who just wanted tips on how to make more side income?