Amid serious concerns about the safety and appropriateness of using xAI’s Grok chatbot within the U.S. government, the U.S. Department of Agriculture (USDA) tells Fast Company that it’s “proud” to move forward with a new plan to use the chatbot at the agency for a range of applications.
The agency’s embrace of Grok marks a major win for xAI, whose chatbot has been plagued by scandal. Last year, the Trump administration announced a series of agreements with major AI companies, including xAI, to make top large language models available to government users at steep discounts. But as officials have moved to adopt models from Gemini and ChatGPT, many have remained wary of deploying Grok. The chatbot raised alarms last year after declaring itself MechaHitler and posting antisemitic responses on X. In January, users generated millions of nonconsensual nude images with the tool, again sparking outcry.
The company made changes to the chatbot in response to both incidents, but federal agencies have remained cautious. As Fast Company reported in January, the General Services Administration has not yet integrated Grok into a government-wide AI tool because it has so far not passed internal safety reviews. The Wall Street Journal also reported in March that Grok had failed government safety evaluations, and federal leaders remained concerned it was too easy to manipulate and overly sycophantic. Federal agencies have shown little interest in adopting the public-sector version, Grok for Government, even as leading members of the Trump administration maintain close ties with xAI CEO Elon Musk.
Now, though, the USDA has decided to move forward with a plan to deploy Grok in its own systems. The agency is beginning that work by sponsoring Grok for review through its FedRAMP program, which essentially amounts to participating in pricey security reviews required before software can be deployed on government cloud systems.
“The U.S. Department of Agriculture is proud to sponsor Grok for FedRAMP authorization to equip our workforce with the most capable AI available and ensure fair competition among providers,” a spokesperson for the agency tells Fast Company. “Grok will undergo the identical rigorous FedRAMP security, privacy, compliance, and responsible-use testing required of every AI provider,” the spokesperson added. “There is no special treatment.” (Fast Company has reached out to xAI for comment.)
Grok for Government was first announced last year, a few days after FedScoop reported that GSA software coders had been working on integrating the software into a government AI resource. As a result of this change, Grok for Government is now listed in an online marketplace for systems undergoing government security reviews. Notably, though, this isn’t the first time the USDA has expressed interest in Grok. Earlier this year, a nutrition website run by the department briefly referenced Grok, before the mention of the xAI tool was removed.
It’s not clear why the Agriculture Department took up the mantle of bringing Grok even further into the government, but the agency handles far less sensitive data than some of its peers, like the State Department and the Department of Homeland Security.
“Grok will be available as an optional tool on the same basis as Copilot and OpenAI models for data analysis, scientific research, conservation planning, agricultural modeling, operational efficiency, and anything that trained internal USDA employees see fit,” the USDA spokesperson adds.