In recent months, officials at several U.S. federal agencies have raised concerns about the safety and reliability of Elon Musk’s xAI’s (X.AI) AI tools, showing continuing disagreements within the U.S. government about which AI models to deploy, The Wall Street Journal reported, citing people with knowledge of the matter.
The warnings came before the Pentagon’s decision this week to allow xAI’s AI chatbot Grok to be used in classified settings, the report added.
xAI did not immediately respond to a request for comment from Seeking Alpha.
Anthropic (ANTHRO) was the only developer approved for classified use before the deal between xAI and the military. However, the Pentagon and Dario Amodei-led Anthropic have been at odds over the use of AI. On Thursday, Amodei rejected the Department of War’s demand for unrestricted access to its AI models. Pentagon leaders, including U.S. Secretary of Defense Pete Hegseth, wanted Anthropic to loosen restrictions on the use of AI.
In recent months, Ed Forst, the top official at the General Services Administration, or GSA, a procurement arm of the U.S. government, raised concerns with White House officials about potential safety issues with Grok, the report noted.
Other GSA officials under him had also shown safety concerns about Grok, which they viewed as sycophantic and too susceptible to manipulation or corruption by faulty or biased data — creating a potential system risk, according to the report.
At the time, in late December and early January, Grok was under scrutiny for allowing sexualized editing of photos, including of children. Government officials saw the issue as representative of how Grok may be exploited by bad actors, the report added.
The matter reached White House Chief of Staff Susie Wiles, who called a senior xAI executive about the concerns. The executive told her that xAI was working on addressing the safety issues that made Grok over-compliant, the report noted.
Josh Gruenbaum, a senior GSA acquisitions official recruited via the Department of Government Efficiency, assured government officials that the government platform of Grok was separate from the public one. Wiles was satisfied, as per the report.
Last month, xAI said it implemented technological measures to prevent Grok from editing images of real people in revealing clothing, and the restriction will apply to all users, including paid subscribers.
In recent weeks, GSA officials were told to put xAI’s logo on a tool called USAi, which is a sandbox for federal employees to experiment with different AI models. Grok has not been made accessible through USAi largely because of safety concerns, and it remains off the platform, the report noted.
The website shows xAI’s logo but only offers models from Anthropic, Google (GOOG) (GOOGL), and Meta (META), as per the report.
A team at the GSA studying AI has circulated a report to top agency officials that indicated Grok’s safety problems. Grok has been suspended for use by GSA staffers for months. Demand from other agencies to use Grok has been weak except in a few cases where people wanted to use it to mimic a bad actor for defensive testing, the report noted.
In a statement, Gruenbaum said the agency takes AI safety seriously. “We rigorously evaluate frontier AI models, including xAI, through a comprehensive internal review process. In this instance, we followed established procedures and maintain our determination to keep it on schedule,” said Gruenbaum.
Two weeks ago, Matthew Johnson, the Pentagon’s chief of responsible AI, stepped down in part over his concerns that safety and governance had become an afterthought amid the Defense Department’s push to expand AI capabilities, the report noted.
Previously, Johnson’s team had circulated memos that highlighted Grok’s safety issues and questioned if it was aligned with government ethics and standards. Those notes had been sent up their chain of command at the Pentagon, the report added.
“The department will continue to ensure the safety and security of all models utilized regardless of classification level,” said a Pentagon official, the report noted.
Earlier this month, Elon Musk’s space company SpaceX (SPACE) announced it was acquiring xAI (X.AI).