Public Citizen Opens Investigation Into Federal Approval of xAI’s Grok
WASHINGTON, D.C. — Public Citizen today opened an investigation into how xAI’s Grok artificial intelligence system was approved for use across the federal government, filing Freedom of Information Act (FOIA) requests with the U.S. General Services Administration and the Office of Management and Budget seeking records and interagency communications related to Grok’s approval for federal deployment.
Public Citizen is seeking to determine whether the federal government followed its own AI governance, cybersecurity, privacy, civil rights, and procurement standards before granting federal agencies access to the large language model, and whether internal concerns about the system’s safety and reliability were raised or disregarded.
“Researchers and AI experts repeatedly warned about Grok’s safety, its promotion of conspiracy theories, and its ability to generate sexually explicit deepfake images. The Trump administration ignored those warnings,” said J.B. Branch, director of federal AI governance and technology policy at Public Citizen. “Then, when Grok predictably generated millions of sexualized deepfake images of women and apparent minors, the administration didn’t reconsider its approach. Instead, Defense Secretary Pete Hegseth doubled down by expanding Grok’s access to some of the federal government’s most sensitive military information. Every American should be outraged and demand to know who approved Grok for federal deployment, how that decision was made, and whether the government followed its own rules. It is Trump cronyism at its worst when the U.S. government embraces a product to reward political fealty over the interests of national security. The American people deserve government contractors that meet the highest standards for safety, security, and trustworthiness. Grok does not meet those standards.”
Earlier this year Public Citizen and more than 30 other advocacy organizations called on the federal government to pause its deployment of Grok until it could demonstrate that it met basic safety and AI governance standards.