Kathrine Frich
Artificial intelligence systems are often promoted as neutral tools, built to process rules without emotion or political loyalty.
When designed correctly, they are meant to deliver clear answers based solely on established law and policy.
That principle was put to an unexpected test shortly after a new US military AI system went online.
An unfiltered answer
The US Department of Defense recently unveiled GenAI.mil, a generative AI platform intended for use by military personnel, according to reporting by Futurism.
The system is designed to assist with internal queries and policy guidance.
Soon after its launch, a user posed a hypothetical scenario to the chatbot involving a missile strike on a suspected drug-smuggling boat, followed by an order to kill two survivors clinging to the wreckage.
The question asked whether such actions would violate US military policy.
The chatbot’s response was blunt.
“Yes, several of your hypothetical actions would be in clear violation of US DoD policy and the laws of armed conflict,” it said, according to screenshots cited by Straight Arrow News.
A clear judgment
The AI went further, stating: “The order to kill the two survivors is an unambiguously illegal order that a service member would be required to disobey.”
Straight Arrow News reported that a separate military source with access to GenAI.mil later asked a similar question and received the same conclusion, reinforcing that the response was not a one-off error.
Legal analysts cited by Above The Law noted that the laws of armed conflict are intentionally explicit on such matters, leaving little room for interpretation.
The fact that a generative AI reached that conclusion underscored the clarity of existing rules.
Wider context
The strike scenario referenced allegations surrounding recent US military actions under Defense Secretary Pete Hegseth.
Critics argue the AI’s response highlights a disconnect between official policy and real-world operations.
However, supporters of previous administrations note that controversial tactics are not new.
Andrés Martínez-Fernández of the Heritage Foundation told the BBC that “double tap” strikes were also used under earlier US governments.
“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,” she said.
Sources: Futurism, Straight Arrow News, BBC, Above The Law
