Grok for Government appears stalled for now, but GSA says current AI listings aren’t final

Elon Musk’s Grok appeared to be under consideration for a major new General Services Administration deal, but the company’s tech is still under review — even as the agency moves forward with agreements with companies like Anthropic and OpenAI. 

Despite unveiling a new Grok for Government product earlier this year, and announcing its inclusion on GSA’s Multiple Award Schedule, xAI’s large language model tech wasn’t included in the rollout of a new governmentwide AI platform called USAi this week. And unlike companies like Anthropic and OpenAI, xAI hasn’t announced a major partnership with the GSA — though the agency maintains that it is considering all AI companies “equally” for government contracts and providing “consistent communication” to every provider. 

Government IT reseller Carahsoft still lists a partnership with xAI on its website, but an archived version of that page previously included xAI technology as part of its federal offerings through GSA. At publication time, the company only listed a state and local offering. In July, a GitHub repository referencing the agency’s work with Grok was also removed from public view after FedScoop asked GSA about its use of the chatbot.

A reference to xAI’s government procurement contract, through Carahsoft, is no longer listed on the reseller’s website.

xAI was apparently supposed to land a deal similar to recent GSA partnerships with Anthropic and OpenAI, but the plan fell through after its chatbot started spewing antisemitic rhetoric, Wired reported Thursday, citing sources with knowledge of the discussions. Those deals appear to be designed to motivate federal agencies to move toward authorizing the technology more quickly, sources told FedScoop earlier this week, and GSA is looking at fast-tracking the FedRAMP process for companies participating in the deals.

One source confirmed to FedScoop there was a June 4 meeting between GSA employees and xAI representatives, adding that Elon Musk did not attend. FedScoop reviewed an xAI enterprise service agreement on Carahsoft’s website with a file that includes the phrasing GSA Approved 6.26.25. The document appears to include modifications sometimes made for enterprise government customers, including an indemnification clause designated as “reserved.” It also references GSA’s Multiple Award Schedule in a data processing addendum. 

A GSA spokesperson provided the following statement:

GSA is fully committed to responsible AI adoption across the federal government, in direct alignment with President Trump’s AI Executive Order and Action Plan.

We are moving quickly, but deliberately, to evaluate a broad range of AI models. Every evaluation follows rigorous safety protocols that prioritize data security and accuracy. This process is ongoing and adaptive. The absence of a particular model should not be interpreted as exclusion or a final determination.

GSA is engaging all companies equally, applying the same due diligence, transparency, and consistent communication to every provider. Approvals are paced based on both provider readiness, our uncompromising evaluation standards, and the need to revolutionize our federal workforce technology.

This careful balance of speed and rigor reflects GSA’s leadership in AI safety and our responsibility to the American public. Our mission is to ensure that every AI tool deployed by the federal government meets the highest standards of security, reliability, and trustworthiness.

The Wired report cited two sources who said they believed the original xAI Grok deal was dropped because the chatbot espouses myriad hateful and antisemitic remarks. Those posts included the chatbot declaring itself “MechaHitler” and conspiratorial commentary about people with Jewish last names. xAI later apologized and blamed the comments on a code path update. Several Democratic members of Congress also criticized the government for working with the company. 

Notably, officials at GSA, including Zach Whitman, the agency’s chief AI officer, have previously told FedScoop that Grok models are being analyzed by a safety team and could be included on its platform in the future. 

“We approve model families based on their passing of these evaluation sets,” Whitman said earlier this month. “Once we evaluate Grok 3 and 4 together, we’ll be able to take that to the safety board. [We can go to them and ask], ‘what do you think about this model family? Is it meeting your standards or not in our behavior?’ So really, it’s just, like, a measurement perspective.” 

The confusion over Grok’s work with the GSA follows clashes between Musk, the Trump administration, and the president himself. There remain ongoing concerns about increasing government dependence on SpaceX, another Musk company, and Starlink, its satellite internet service. 

Neither xAI nor Carahsoft provided a comment by publication time. 

An earlier version of this piece included a picture of the current xAI state and local offering. This piece now includes an image of the federal offering for xAI.