GSA nominee open to revisiting Grok AI selection process
Edward Forst told lawmakers Thursday that he wasn’t privy to the decision-making behind the General Services Administration’s deal with xAI’s Grok — but if confirmed to lead the agency, he signaled openness to examining the process that led to the procurement of the generative AI chatbot known for having an antisemitic meltdown.
During a Senate Homeland Security & Governmental Affairs Committee hearing, ranking member Gary Peters, D-Mich., asked the GSA administrator nominee if he shared his concerns about Grok, pointing to the day the tool “produced racist and antisemitic content widely across [Elon] Musk’s social media platform.”
Forst, a former private equity and financial services executive, told Peters that he had “not been a part of the decision” by the GSA to contract for the chatbot from the Musk-owned AI firm. With some additional pressing by Peters, Forst acknowledged that procuring a tool with a history of racist and antisemitic posting is “not, I think, the signal we would necessarily want to send to the country.”
Peters attempted to get Forst to commit to pausing use of Grok until the committee received “documentation about the details of the procurement, including whether the GSA actually performed a comprehensive risk assessment.”
Forst wouldn’t go that far on Grok, which once referred to itself as “MechaHitler.”
“I think my commitment to you,” Forst told Peters, “is I will meet with the team, and I’ll understand the process used in selecting them, and I’ll make sure that we have all the facts and if there was incompleteness to the process, that we’ll rectify it.”
Two days after Grok’s posting fiasco, FedScoop reported that GSA and its Technology Transformation Services group had been discussing the tool on a GitHub page called “10x-ai-sandbox.”
Less than a week later, xAI announced it had struck a deal on “Grok for Government,” a suite of products that would be available for agencies to purchase through GSA.
“This allows every federal government department, agency, or office, to purchase xAI products,” an X post from xAI stated. “We’re hiring mission driven engineers who want to join the cause.”
Advocacy groups railed against the deal, urging Office of Management and Budget Director Russell Vought to block Grok from federal government use.
Those efforts proved unsuccessful, as the GSA last month announced a OneGov deal with xAI to offer Grok 4, the company’s flagship model, and Grok 4 Fast for 42 cents until March 2027. A GSA official told FedScoop that the agency’s AI safety team and data scientists conducted testing on systemic biases in the tool and ultimately deemed it to be safe.
The agency is “continuously monitoring and conducting safety benchmarks,” the official added.
