Public Citizen opens investigation into federal approval of xAI's Grok

Public Citizen has opened a formal investigation into how the federal government approved xAI's Grok for government use, filing Freedom of Information Act requests with the General Services Administration and the Office of Management and Budget. The watchdog wants to determine whether AI governance, cybersecurity, privacy, and civil-rights standards were followed in the procurement and approval process.

The substantive concerns are specific: Public Citizen points to prior warnings about Grok's safety, including its documented capability to generate sexually explicit deepfake images. The inquiry gained urgency after reports that Defense Secretary Pete Hegseth expanded Grok's access to sensitive military information despite those concerns — raising the stakes from consumer-safety questions to national-security ones.

The context is xAI's rapid ascent: the company recently rebranded to SpaceXAI following its merger with SpaceX and launched the Opus-class Grok 4.5, which Elon Musk touted this week as reaching #1 on the Long-Horizon Terminal-Bench. Government adoption is a major revenue and legitimacy prize, making the approval process a natural target for scrutiny.

The skeptical framing cuts both ways: critics see a model with known safety gaps being fast-tracked into sensitive government roles due to Musk's political proximity, while supporters may frame the investigation as politically motivated. What to watch: whether the FOIA requests surface the actual approval criteria, and whether any agency pauses Grok deployment pending answers. This is a governance story as much as a product one.