Grok Sees Minimal Adoption in US Government Records

According to a Reuters review of federal inventory records, Grok, the chatbot from Elon Musk's xAI, appears in only 3 of more than 400 publicly identified examples of federal AI use during 2025. Reuters reports that those three entries covered basic tasks, two routine administrative uses at the Election Assistance Commission and one Department of Energy pilot at Lawrence Livermore National Laboratory for document summaries and research assistance. By contrast, Reuters found 234 instances listing technology based on OpenAI's models, with Google and Anthropic appearing in dozens of entries. Reuters also reports that Grok has been offered to federal agencies for about $0.42 per agency and that the Office of Management and Budget did not respond to Reuters' requests for comment.

Wired reports that SpaceX's IPO filing disclosed potential legal and reputational risks tied to Grok's permissive features, noting the filing listed Grok's "Spicy" and "Unhinged" modes as exposures and reported that the company set aside about $530 million for potential litigation losses related in part to complaints over sexualized imagery. Bloomberg reports that xAI offered employees $420 to submit completed tax returns for training Grok ahead of the U.S. tax deadline and that payments had not been made as of the Bloomberg report, citing internal chats and company communications. Reuters additionally notes that xAI did not respond to detailed questions about Grok's government usage.

Observed patterns in similar deployments show that government adoption metrics in public inventories are a noisy, conservative signal of enterprise traction. Public federal records often capture pilot projects and vendor mentions unevenly, and many agency procurements do not name specific third-party models in their public filings. For practitioners, low appearance counts in these records generally indicate either limited agency uptake, narrow pilot scopes, or procurement choices that conceal vendor-level detail rather than definitive proof of technical incapability.

Industry observers note that chatbots marketed on openness or permissive output modes frequently face higher safety, procurement, and regulatory friction. Reporting by Wired and Reuters places Grok's permissive features and related complaints in the same frame as financial and legal risk disclosures in SpaceX's IPO paperwork. Public-sector buyers and enterprise customers tend to prioritize auditability, safety controls, and vendor support, attributes that are often decisive in procurement even when list pricing is nominal.

• •Adoption signals: filings or contracts that name specific vendors, particularly GSA schedules or agency procurement notices, will clarify whether Grok advances beyond small pilots.
• •Regulatory and legal outcomes: progress in the investigations and class actions mentioned in Wired's coverage and any enforcement actions could affect market access.
• •Product changes and guardrails: announcements or changelogs from xAI that document moderation, safety layers, or enterprise features would be an observable response that enterprise buyers monitor.
• •Employee and data-practice issues: reporting such as Bloomberg's account of unpaid incentives and internal data-collection practices is relevant to compliance and model-audit concerns.

The collected reporting from Reuters, Wired, and Bloomberg documents weak public evidence of federal adoption for Grok in 2025, contemporaneous legal and reputational disclosures tied to the product, and internal data-practice questions at xAI. Industry-pattern observations suggest these factors are material to procurement and enterprise trust, but public inventory counts alone do not fully reveal the technical performance or all procurement pathways through which a vendor might gain traction.