No longer even a “MechaHitler” meltdown used to be ample to fright off the Pentagon, curiously.
Precise days after xAI’s Grok chatbot generated a circulation of antisemitic and homophobic responses, Elon Musk’s company offered that it secured a $200 million protection contract to originate and offer AI tools to federal workers.
The Department of Defense’s Chief Digital and Artificial Intelligence Location of business offered the contract on Monday, announcing it is miles designed to help the authorities faucet into primarily the most up-to-date AI tools for every little thing from battlefield operations to industry wants.
“The adoption of AI is reworking the Department’s potential to offer a boost to our warfighters and reduction strategic profit over our adversaries,” CDAO Chief Digital and AI Officer Dr. Doug Matty stated in a press beginning. “Leveraging commercially accessible choices into an built-in capabilities manner will velocity up the usage of superior AI as segment of our joint mission an necessary duties in our warfighting domain apart from to intelligence, industry, and project files programs.”
In retaining with Lucas Hansen, the co-founding father of the AI security and training nonprofit CivAI, the deal with xAI is less about quick deny and more about retaining inch with global rivals.
“The authorities desires to offer a boost to the frontier labs and pattern of AI in deliver that the U.S. doesn’t tumble within the help of China,” Hansen informed Decrypt. “Then there’s the quiz of how those funds are disbursed and how that give a boost to is geared up. I deem this could maybe maybe maybe upright be the particular manner they’ve chosen to offer a boost to the firms, as a substitute of an expectation of internal deployment straight.”
Nonetheless Hansen warned that Grok’s underlying habits raises deeper issues about the draw in which it will behave in excessive-stakes authorities settings.
“Even supposing Grok is deployed by the authorities utilizing the an identical model that beforehand confirmed a willingness to generate depraved responses, then—even if those deny cases aren’t explicitly antisemitic or hateful—it’s primarily more at possibility of behave unethically and autonomously,” he stated.
Grok has been the center of several public meltdowns. In Can also, reports surfaced that Grok inserted “white genocide” claims into unrelated prompts. xAI blamed the mission on a rogue worker. Final week, upright old to the Grok 4 beginning, the chatbot again drew criticism for a wave of racist and homophobic outputs, ensuing within the resignation of X CEO Linda Yaccarino.
Whereas issues linger over how Grok could maybe maybe behave in quiet environments, Hansen stated the model the Pentagon uses is unlikely to be the an identical as the one who X users score had regain admission to to since November 2023.
The $200 million award is truly the most up-to-date deal between the United States navy and high AI builders. Other AI builders receiving $200 million contracts include Google, Anthropic, and OpenAI. In November, Meta stated it would build its beginning-source Llama AI model accessible to U.S. protection agencies and contractors. In June, ChatGPT developer OpenAI used to be awarded its have $200 million contract with the Department of Defense, launching its AI suite, OpenAI for Executive.
If the Pentagon deployed Grok over a rival, Hansen emphasized it’d be because Grok is “beginning-weight,” that manner it ought to also be flee on deepest hardware and no longer require an API connection help to xAI.
Asserting Grok for Executive – a series of merchandise that build our frontier items accessible to United States Executive prospects
We’re particularly appealing on two unique partnerships for our US Executive partners
1) a unique contract from the US Department of Defense
2) our…— xAI (@xai) July 14, 2025
Despite Grok’s most up-to-date controversies, on the center of the deal, Hansen stated, will seemingly be a easy strategic calculation: End finish to the important thing avid gamers shaping the vogue forward for AI.
“I deem most of right here’s the authorities making an strive to score a relationship with the whole frontier labs, and apart from they hit the four labs that in actuality depend,” he stated.
In a separate commentary announcing the deal with the Pentagon, xAI stated its mission is to manufacture AI tools to “help humanity in our quest for understanding and data.” To halt this, xAI offered the beginning of Grok for Executive, a series of AI tools including the currently launched Grok 4 and Deep Search.
“The USA is the enviornment leader in AI, and right here’s in no little segment ensuing from a convention of innovation and stable investments in engineering and science,” xAI wrote. “We’re wrathful to make a contribution help to the nation that made xAI uniquely conceivable right here.”
xAI did no longer acknowledge to requests for feedback from Decrypt.
