The Pentagon wants Terminators and it isn’t even funny
June 29, 2026
By Matthew Pietz
This article was written and edited without the use of AI.
There is no rational explanation for US Secretary of Defense Pete Hegseth’s drive to create fully autonomous, lethal robots. At Keranaut, we've searched high and low for a rationale and even spoken to an insider in the US Army robotics program, and we conclude he just wants the US military to be as deadly as possible and does not give weight to any other factor. This tunnel vision could have dramatic consequences for the world.
Lethal machines are already a central part of warfare (80% of the strikes by both Russia and Ukraine are via drone), and for good reasons. Putting a machine at the front saves human lives. It’s the desire to let those machines choose and fire on targets all by themselves that should terrify us.
A Pentagon memorandum on AI claims “the risks of not moving fast enough outweigh the risks of imperfect alignment”. Since “alignment” means AIs and humans wanting the same thing, “imperfect alignment” here means robots killing people they were not told to kill. And Hegseth is not alone: sober defense analysts and non-partisan think tanks also argue for more autonomy in war machines, citing, for example, the waste involved in having six people operate and approve decisions of a single drone.
Are “waste” or “not moving fast enough” really the greatest problems we face?
At Keranaut we strongly feel the greater evil is opening the Pandora’s box of autonomous killing machines. (Though technically, it is already open. This month a Ukrainian defense contractor claimed the first fully autonomous robot kills already happened, two years ago.)
The reasons seem too obvious to bother listing, but not everyone is convinced. So:
Anyone who has asked ChatGPT a fact about the world and gotten back a total fabrication can imagine how an AI could hallucinate that a civilian is a legitimate target, that a hospital is an army barracks. It would just be a matter of time.
A small number of hackers could gain control of many war machines. Certainly, the Pentagon would have encryption, but Russia has successfully gotten malware into Pentagon systems (SolarWinds, 2020), and China hacked into 400 government departments in 2025 including the National Nuclear Security Agency. What target could be more attractive to hackers, government or independent, than robot soldiers?
Someday, AI may well be smarter than us. Even if there is a small chance, we should do what we can to make sure it isn’t holding guns when that happens.
We spoke to a logistics expert in the US Army’s robotics division who says remote-control robot programs are being shuttered as “autonomy” becomes the new buzzword in the halls of the Pentago. “It’s clear that Hegseth wants the ability to use robots and AI to wage unlimited lethal war without guardrails or limitations,” he says. “This is perfectly in character with the ‘Epic Fury’ type of bro-ism that this former Fox News celebrity exemplifies. He’s also an evangelical Christian, and I don’t know how robots and AI fit into that religious worldview, but it might be why he’s not afraid of creating a world-destroying AI.”
Congress has tried to rein in Secretary Hegseth’s impulses. This month Sen. Kirsten Gillibrand introduced a tough bill to force humans into the loop for lethal targeting or nuclear launches, and she hopes to get it incorporated into the National Defense Authorization Act; this is better than Sen. Elizabeth Slotkin’s similar bill which grants the executive branch waivers so they can use lethal AI without getting anyone else’s buy-in.
The White House countered these reasonable steps in June with an executive order stating AI should be “deployed rapidly to confront any and all threats.” Threats that are worse, presumably, than the threat to humanity from leaders with an adolescent fascination with killer robots.
Write your Senator to support Gillibrand’s bill or donate to a non-profit working on this critical issue. If we can’t trust AIs to count R’s in “strawberry”, we can’t trust them with tanks. It sounds stunningly obvious, but people with extraordinary power over our future do not seem to get it.
Click here to subscribe and be notified of future posts.
