The United States is trying to do two things at once with artificial intelligence in national security: move faster to adopt it, and make sure a person remains answerable for what it does.
In an opinion piece published by The Hill (and carried on MSN), the argument is that the promise of "agentic" AI — systems that can act on their own to complete tasks — is enormous, but that America cannot deploy it at scale today. The reason, the authors write, is reliability: current systems simply aren't dependable enough yet to be trusted with government work. Their conclusion is that building that reliability is the precondition for unlocking AI's value.
While technologists push on reliability, lawmakers are focused on responsibility. According to Small Wars Journal, Sen. Mark Kelly has introduced an amendment to the defense bill that would ensure "ultimate human responsibility" in any AI-powered "kill chain" — the sequence of steps that leads to a weapon being used. The language is meant to keep a human accountable for lethal decisions rather than ceding them to software.
The push is not universally welcomed. ZNetwork reports that many anti-imperialist, antiwar and Palestinian solidarity organizers in Boston are examining the city's connections to the development of AI weapons technology, reflecting activist unease about where these tools come from and how they will be used.
Together, the items sketch a single moment: enthusiasm for military and government AI, matched by open questions about whether the technology is trustworthy and who bears the blame when it fails.
Why it matters: as AI moves toward decisions that can cost lives, the fight over keeping a human clearly accountable will shape both how quickly these systems are fielded and how much the public trusts them.