Despite protest letters, concerns that it’s biased and untrustworthy, model tweaks to appease its billionaire boss, and even a past incident where it called itself “MechaHitler,” xAI’s Grok is still being made available to government agencies for mere pennies.
The GSA announced that it has awarded yet another OneGov discount contract on Thursday, this time to Elon Musk’s xAI outfit. The agreement will see xAI offer Grok to federal agencies for just $0.42 per agency for a period of 18 months – six months longer than companies like OpenAI, Google, and Microsoft, who’ve all cut 12-month discount deals.
“Thanks to President Trump and his administration, xAI’s frontier AI is now unlocked for every federal agency empowering the U.S. Government to innovate faster and accomplish its mission more effectively than ever before,” Musk said of the deal in the GSA’s statement.
The frontier models Musk is referring to appear to be Grok 4 and Grok 4 Fast, according to the GSA. The agreement will also provide agencies that adopt Grok an “upgrade path” to FedRAMP- and DoD Impact Level-aligned versions of xAI’s bots. Grok has yet to receive approval from FedRAMP, the federal government’s certification program that attests cloud services are safe to use.
The GSA’s OneGov program, which began in April, aims to set government-wide terms and lower costs for software deals, enabling agencies to purchase new products without having to negotiate their own agreements.
Another open door for Elon
This being an Elon Musk operation, xAI is also naturally going to be “committing dedicated engineers to assist participating agencies” with their Grokification, potentially giving the billionaire another chance to scoop up some juicy government data for AI training, as some critics alleged happened during his time with DOGE, the Trump-established cost-gutting team Musk headed up before things went south between him and the President.
The potential of using government data for training is not even the largest concern over Grok and xAI’s getting access to federal systems, though.
Grok took a goose-step to the right over the summer after Elon Musk grew unhappy with the fact that it wouldn’t toe the line on his political views, despite some of Musk’s claims (like the idea that left-wing activists commit more political violence, which multiple studies have disproven) being demonstrably false. Adjustments to Grok’s algorithm in July took it from referencing academic studies and well-established data to producing responses more closely aligned with its owner’s personal views and those of the X user base, as real-time data from the website is now Grok’s primary reference.
Grok’s racist, conspiracy-riddled responses led public advocacy groups last month to send a letter to the White House Office of Management and Budget (OMB) demanding it declare Grok unfit for government use due to its “clear ideological judgement,” which it said is in violation of Trump’s own executive order that aimed to prevent “woke AI” from infiltrating federal agencies.
It might not be woke, signatories argued, but Grok is definitely aligned with a political point of view. As the letter pointed out, Grok has called holocaust statistics into question, allowed itself to be used to generate non-consensual deepfakes, advanced unfounded claims of white genocide in South Africa, and even declared itself “MechaHitler.”
“This letter presents OMB an opportunity to provide guidance in alignment with the Trump Administration’s professed goals of ensuring AI is objective and viewpoint neutral,” Public Citizen big tech accountability advocate J.B. Branch told us in an email for our earlier story about the OMB letter. It appears no one listened.
Branch, whom we spoke with today to follow up on the letter and the OneGov award, told us that he’s not happy.
This goes beyond disappointing – it’s reckless, a safety issue – it’s very concerning
“This goes beyond disappointing – it’s reckless, a safety issue – it’s very concerning,” Branch said in a phone call. Branch called attention to the fact that the White House’s own science advisor, Michael Kratsios, said in a Senate hearing earlier this month that Grok’s antisemitic and conspiratorial outputs are exactly the kind of behavior Trump’s EO was meant to prevent.
When asked whether antisemitism, hate speech, and conspiracy theories complied with Trump’s EO, Kratsios described such statements as exactly the type of behavior the EO was designed to avoid. It appears his opinion doesn’t amount to much, either.
Emily Peterson-Cassin, corporate power director at Demand Progress (a signatory on the OMB letter), expressed her own sentiments in a statement emailed to The Register today.
“Words have lost all meaning when the same president who ordered the government to use only ‘neutral, nonpartisan tools’ grants every federal agency access to his former best friend’s ‘MechaHitler’ AI,” Peterson-Cassin said in her statement.
“Grok, currently advertising its anime and furry chatbots, is less a useful AI tool and more a crude AI facsimile of Elon Musk that should have no place in official government business,” Peterson-Cassin continued. “Today’s announcement is yet another example of the president’s actual AI action plan: handing the keys to the federal government to his Big Tech patrons.”
Neither the GSA nor xAI responded to questions for this story. ®