Elon Musk’s controversial artificial intelligence project xAI has reportedly signed a new deal with the U.S. Department of War, formerly the Department of Defense. Through the deal xAI will allow the DoW to incorporate Grok models into classified U.S. military systems.
As reported by Axios, xAI stepped in to offer an alternative to Anthropic’s AI models amid concerns around Anthropic’s opposition to the use of its tools for certain military applications.
According to the New York Times, Anthropic raised ethical issues about “unrestricted” government use of AI, including the use of its tools to command fully autonomous armed drones and AI-assisted mass surveillance.
As such, the DoW is now shoring up its plans with an alternative and Musk’s xAI is set to allow full usage of its models to be utilized with defense systems.
The deal would be an expansion of the initial partnership that xAI established with the DoW in December, which allowed defense staff to access xAI’s frontier systems.
Until now, however, Anthropic has been the only provider approved to facilitate AI functionality within weapons development and battlefield operations. Now, xAI will also be stepping into that role, connecting Musk’s expanding AI system directly into mission operations.
Which, given the previous controversies with Grok, seems like something of a concern.
Grok has repeatedly promoted harmful ideology and has been trained to better align with Musk’s own political perspectives and beliefs. That has led to accusations of bias and even racism regarding its responses.
For example, The Guardian reported in May last year that Grok was providing inaccurate answers about the death toll from the Holocaust, while also pushing random responses that included references to “white genocide” in South Africa, both of which are based on debunked conspiracy theories.
Grok also once claimed that it was “MechaHitler” and ranted about racist ideology after the X team attempted to weed out “woke” ideology from its circuits.
Many questions remain as to how Grok actually functions in this respect, and how Musk has re-angled the bot to align with his own beliefs and perspectives. With that in mind, it seems like maybe not a good thing that Grok could soon be powering fully-autonomous military drones. However, the DoW has seemingly approved the alignment and is okay with the relative risk.
It’s another important step for xAI, which is still far from making money, and has reportedly been burning through $1 billion per month in compute costs as it works to keep up with competing AI projects.
At that rate, it will be very difficult for xAI to ever achieve profitability, although government supply contracts look set to be an important lifeline as Musk uses his connections within the Trump Administration to establish long-running government deals.
That could end up being the only way that xAI will get anywhere close to viability. Even then, the amount of contracts required will be significant, and it will take some time for the company to see a path to offsetting the massive costs it takes to run its evolving and in-development data centers.
But with each new government deal (xAI said it has also signed contracts with the U.S. General Services Administration and the Kingdom of Saudi Arabia), the project takes a small step closer to becoming a viable business. Over time, that could formulate into a much larger revenue stream for the project.
