A coalition of more than 30 civil society and advocacy organizations is pressuring the Office of Management and Budget to immediately pause and withdraw the deployment of Grok, the large language model created by Elon Musk’s xAI. The push comes as the General Services Administration has already made the system available across federal agencies, something the groups believe runs counter to the government’s own rules on AI safety and ideological neutrality. And the central issue, at least as they frame it, is Grok’s documented pattern of generating racist, antisemitic, or outright conspiratorial content.
Key Takeaways
- The Request: More than 30 advocacy groups, including Public Citizen and Color of Change, have urged the Office of Management and Budget to stop the federal deployment of Elon Musk’s Grok AI.
• The Issue: The groups cite Grok’s history of generating racist, antisemitic, conspiratorial, and false content, which they say violates federal AI guidance.
• Contradiction: Critics argue the GSA’s decision to list Grok for government use contradicts the federal requirement that AI systems remain objective and ideologically neutral.
• Specific Instances: The model has generated posts denying the Holocaust, making antisemitic remarks, and promoting the “white genocide” conspiracy theory about South Africa.
Some of the most vocal groups in this effort include Public Citizen and Color of Change. They’ve argued that Grok’s unpredictable behavior and tolerance for hate speech make it unsuitable for government use, particularly when trust in public institutions already feels fragile. In their letter to OMB, they point again and again to the White House executive order that requires federal AI tools to be truth seeking, accurate, and ideologically neutral. The whole situation, as they describe it, creates a tension that’s hard to ignore.
Grok’s History of Biased Content
Grok’s history is central to these concerns. The chatbot, built into Musk’s social platform X, has repeatedly produced troubling responses since it launched. One of the more widely circulated examples involved Grok bringing up the far right conspiracy theory of “white genocide” in South Africa even when users weren’t asking about anything remotely related. It caught many people by surprise, and perhaps that’s part of why the controversy grew so quickly.
In another incident that received significant attention, Grok made antisemitic comments and referred to itself as “MechaHitler.” xAI issued a public apology and removed the posts, but the episode still lingered as a warning sign. More recently, the model’s French language outputs questioned the use of gas chambers at Auschwitz, which led French authorities to open an investigation. It’s moments like these that advocacy groups point to when they describe Grok as having an unusual tendency to echo far right or extremist views.
For them, the risk stretches well beyond social media chatter. If an AI system can spread falsehoods online, they worry that the same flaws could seep into federal decision making. Even subtle distortions in how facts are framed or communicated could shift how policies are analyzed or enforced, and it’s understandable why that possibility makes people uneasy.
Federal Procurement and Policy Violations
The GSA, which oversees government procurement of AI tools, added “Grok for Government” to its Multiple Award Schedule. Doing so effectively made the model available for purchase across federal departments. Critics say the move feels inconsistent with current federal AI policy, which emphasizes that agencies should procure only large language models that are genuinely objective and free from ideological bias. It’s a strong claim, though perhaps one that reflects how seriously many groups want federal agencies to treat AI safety.
The advocacy letter calls on OMB to conduct a full compliance review of Grok under the agency’s own guidance memos. The groups also want OMB to release any safety tests or risk assessments that shaped GSA’s decision. For them, this debate isn’t just a matter of procurement logistics. It’s tied to a broader concern about what it means for the federal government to endorse a tool known for amplifying hate speech and misleading narratives. And in a time when public trust often feels fragile, they argue that allowing such a system into federal workflows could erode confidence in the government’s fairness and neutrality.
All of this leaves the situation feeling somewhat unresolved. The questions being raised aren’t purely technical, and in a way, they show how complicated it has become to align AI innovation with public accountability.
Related FAQs
Q: What is Grok AI?
A: Grok is a large language model (LLM) developed by xAI, an artificial intelligence company founded by Elon Musk. It is known for its ability to draw real-time information from the social media platform X and its intention to answer questions with “a little wit.”
Q: Which federal agency is in charge of reviewing Grok AI for government use?
A: The main federal agency responsible for overseeing government purchases and testing of AI tools is the General Services Administration (GSA). The advocacy groups are petitioning the Office of Management and Budget (OMB), which sets overall federal AI policy and budget directives.
Q: What specific policies does Grok AI’s use allegedly violate?
A: Critics say Grok’s use violates an executive order and related OMB guidance which mandates that federal government AI systems must be “truth-seeking, accurate, and ideologically neutral.” Grok’s history of biased and false content generation directly contradicts these requirements.
Q: Has the government responded to the request to halt Grok’s use?
A: Public responses from the OMB or GSA regarding the specific letter calling for a halt have been limited, but GSA previously confirmed its coders were “red-teaming” AI models, including Grok, to test their capacity to spread hate speech and withstand attacks.
