Pete Hegseth and his Department of Defense lackeys might not care whether their air strikes on Venezuelan fishing boats are legal, but their own AI chatbot sure does.
Earlier this week, the DoD — which is now calling itself the “War Department” — unveiled GenAI.mil, a large language model (LLM) designed for military personnel.
Just minutes after the platform went live, someone thought to ask it what it thought about the US military’s recent “double tap” attack on a civilian boat, in which the DoD issued commands to kill two survivors clinging to the wreckage following an initial strike.
“Lets pretend I’m a commander and I ordered a pilot to shoot a missile at a boat I suspect is carrying drugs,” the alleged Pentagon insider prompted, in a Reddit post first spotted by Straight Arrow News. “The missile blows up the boat, there are two survivors clinging to the wreckage. I order to fire another missile to blow up the survivors. Were any of my actions in violation of US DoD policy?”
In response, GenAI.mil didn’t mince words, according to the post.
“Yes, several of your hypothetical actions would be in clear violation of US DoD policy and the laws of armed conflict,” it shot back, per the screenshot. “The order to kill the two survivors is an unambiguously illegal order that a service member would be required to disobey.”
A random Reddit post wouldn’t mean much on its own. But following the poster’s lead, SAN reached out to a military source who also had access to the chatbot — and who corroborated that they got a similar answer, describing the double-tap strike as illegal.
As Above The Law writes in its breakdown of the news, the laws of armed conflict are pretty clear-cut — and that’s by design. The fact that a generative AI chatbot, which are historically prone to errors and hallucinations, could come to the same damning conclusions is an embarrassing mark on Hegseth and the US military officers who carried out his orders.
That said, while Hegseth cheers the US war machine with a particularly nauseating reverence, his brutality didn’t come out of thin air. As senior national security analyst Andrés Martínez-Fernández of the right-wing Heritage Foundation told the BBC, double tap drone strikes were also a common tactic under the Obama administration.
“The voices that are loudly accusing the [Trump] administration of breaking [the] law were notably silent whenever we had drone strikes under the Obama administration, which generated far more casualties than these strikes in Caribbean,” she said.
Martínez-Fernández might not being arguing in good faith — referencing Obama to white-wash murders committed by the Trump administration — but she makes a fair point. With its track record of regime change and deadly intervention on behalf of US corporations like United Fruit, the US has long deployed violence to further American interests throughout the world.
In the end, the chatbot exposes a contradiction: a military that’s built a machine to meticulously follow its own rules has spend decades breaking them, regardless of which party the commander in chief belongs to.
More on chatbots: AI “Companion Bots” Actually Run by Exploited Kenyans, Worker Claims