Humans Don’t Kill People. Drones Do.

Have you ever asked yourself “who is getting rich off killer robots?”

Meet Palmer Luckey, founder of the defense-tech startup Anduril, who gave a TED talk where he literally declared “I love killer robots.” At least part of the reason for that is because he’s making tens of billions selling autonomous arms. His first backer was Peter Thiel, whose company Palantir runs the surveillance that helps decide where strikes land. Anduril has raised at least $6.9 billion—though other estimates put the number upwards of $60 billion.

Luckey is not alone. Eric Schmidt, the former Google CEO who championed the now abandoned company motto “Don’t Be Evil,” now runs a drone-warfare operation nested inside a Russian doll of shell companies that supplies the front lines of the Ukrainian war effort. Eric Trump is chief strategy adviser to a startup building humanoid killer robots that just landed a $24 million Pentagon contract. The Department of Defense is pouring tens of billions into overhauling America’s drone programs. For scale, compare the number of drone deaths in Ukraine (124,000/year) to the devastation of the opioid epidemic (33,000 deaths per year). Not only do drones carry a higher annual body count, but the Sackler made a comparable pittance for their effort: only $10 billion from two decades of drug dealing.

Science fiction writers have predicted this exact situation for at least 80 years. In 1942, Isaac Asimov wrote the first law of robotics: a machine may never harm a human being. But the instant we actually had access to automated death machines we abandoned any effort of actually regulating their existence.

Now the cost of armageddon is plummeting.

Today there are two kinds of murderbots: the expensive American kind, and the DIY kind almost anyone can build. Think ISIS strapping grenades to hobby drones a decade ago. Meanwhile a moderately funded tourists state could do exponential levels of harm. AI Frontiers recently ran the numbers on cheap tactical drones: a single box truck could carry 40,000 of them at maybe $25 each. Every one could be given a mission to hunt the first person it sees. The war isn’t only on foreign shores any more. It’s coming to a city near you.

Take for instance the venture-backed anti-school-shooter drone company Campus Guardian Angel. It seems like a reasonable solution to an intractable problem—drones trump guns in almost every engagement. But the real issue is that what starts as a narrow use case has the tendency to expand to others. Do we want police to use lethal drones on American streets? And what happens when they kill the wrong person?

Ask Luckey and his answer mirrors theNRA’s: drones don’t kill people, people do. Convenient, for the guy getting rich off the drones.

Here’s the only good news in the whole story: none of this is settled. The UN set a 2026 deadline for a binding treaty on autonomous weapons, built around one rule — keep a human in the loop. 166 nations wanted it on the table. Three said no. The deadline passed with nothing, because the powers building these weapons didn’t want to be told no.

That law can still be written — but only from the bottom up. So ask your representatives one question and don’t let them dodge it: do you support a binding human-control standard for autonomous weapons — yes or no? Then follow the money to whoever wants the answer to stay no. For more information and an action plan of what you can do next, check out the Center for AI Safety.

The drone war isn’t coming. It’s here. The full investigation is up now on the channel.