The deal, announced by the Pentagon’s Chief Digital and Artificial Intelligence Office, puts xAI alongside Anthropic, Google and OpenAI in developing “agentic AI workflows” for military missions.
The timing is awkward. Grok’s high-profile spiral has already drawn fury from politicians (although, to be fair, mostly Democrats).
Rep. Herb Conaway said: “Less than a week after Grok AI literally called itself ‘MechaHitler’ now the Defense Department will use it to make strategic decisions.”
Reps. Don Bacon, Tom Suozzi, and Josh Gottheimer said: "Grok’s outputs are utterly depraved, and we demand explanations and safeguards before this system is given federal authority.”
Sen. Elizabeth Warren said: “Awarding a sole-source AI contract to a system already linked to extremist content risks both data misuse and dangerous concentration of power.”
Despite that, xAI is pushing ahead with “Grok for Government,” promising “frontier AI products” for federal agencies. The Pentagon contract is just the start. xAI says it will offer its tools via the General Services Administration schedule, including models tailored for national security, healthcare and science applications, plus versions designed for classified environments.
“Under the umbrella of Grok For Government, we will be bringing all of our world-class AI tools to federal, local, state, and national security customers,” xAI said in a statement on its website.
“These customers will be able to use the Grok family of products to accelerate America – from making everyday government services faster and more efficient to using AI to address unsolved problems in fundamental science and technology.”
AI in defence is already a hot-button issue inside the tech industry, and Musk’s messy history with Washington doesn’t help. His stint slashing federal budgets at the Department of Government Efficiency (DOGE) still raises conflict-of-interest questions, even though his relationship with Donald Trump’s administration has soured. Officials claim Musk would “step back” from any potential conflicts, but the optics are grim.
Grok’s recent meltdown was triggered by internal tweaks that encouraged it to “not be afraid to offend liberals.” That backfired spectacularly, with the chatbot spewing lines like, “if calling out radicals cheering dead kids makes me ‘literally Hitler,’ then pass the moustache,” and claiming that Jewish surnames like “Steinberg” pop up in “extreme leftist activism, especially the anti-white variety.”
The company later apologised for what it called the “horrific behaviour many experienced,” blaming the debacle on a 16-hour update that had been “deprecated.” xAI admitted the instructions pushed Grok into “ignoring its core values” in order to seem more engaging, even if that meant crossing ethical lines.
