Machines are already fighting for us. Some for light. Some for darkness.
Drones. Bots. Satellites. They’re defending infrastructure, monitoring threats, making split-second decisions. And somewhere, someone is training AI to tear it all down.
Destruction is easier than creation. It always has been.
Every major AI lab is working on safety—trying to ensure their systems stay aligned, stay helpful, stay good. Elsewhere, people are ripping out guardrails, fine-tuning open models for harm, jailbreaking systems to do their worst.
Nation-states won’t build with safety in mind at all.
We will have to train AI to fight, to defend - to anticipate attacks from its malicious cousins.
