Reports of state-affiliated actors attempting to misuse AI are a reminder that safety is not a future problem. It is already a product responsibility.