Senator Elizabeth Warren is pressing Defense Secretary Pete Hegseth for a full accounting of xAI's access to the Pentagon's classified systems, expressing direct concerns about the security of the company's Grok chatbot. In a letter sent Sunday, the Massachusetts Democrat and member of the Senate Armed Services Committee warned that a potential leak of government information by the AI could expose military plans, intelligence operations, and endanger personnel.

Warren's four-page letter questions what security evaluations the Department of Defense conducted before reportedly granting Grok access. "It is unclear what assurances or documentation xAI has provided," she wrote, asking for the complete contract between the Pentagon and Elon Musk's company.

The inquiry follows a deal struck last month to integrate xAI's models into the military's secure networks. Pentagon spokesperson Sean Parnell defended the partnership, stating, "xAI’s Grok remains a competitive frontier model — no media spin can change that." He confirmed plans to deploy Grok on the Pentagon's official GenAI.mil platform soon.

This agreement was finalized as the Defense Department's separate and public negotiations with Anthropic stalled over that company's restrictions on using its Claude model in weaponry and surveillance. The Wall Street Journal recently reported that officials at several agencies, including the National Security Agency, have separately questioned Grok's reliability.

xAI secured a $200 million contract with the Pentagon last July. Warren's letter asks what specific protections are in place to guard the system against cyberattacks that could corrupt its outputs. The Pentagon has not yet publicly responded to her requests.

Source: The Hill