The government agencies have raised alarm over Elon Musk’s Grok as officials warn that the chatbot is not reliable and safe enough for military use.
According to the Wall Street Journal, the Pentagon is moving forward with integrating Grok into classified and secretive operations. This shift is driven by Anthropic’s firm stand over AI safety rules, refusing to drop these guardrails for unrestrictive military use.
Unfortunately, the Grok chatbot is not free of alarming concerns. A General Services Administration (GSA) report found that “Grok-4 does not meet the safety and alignment expectations required for general federal use.”
The report also labeled it “sycophantic and susceptible to corruption by biased data,” thereby posing elevated and difficult-to-manage safety risks.
A classified November 2024 review by the National Security Agency (NSA) also identified security vulnerabilities in Grok that were not present in other models like Anthropic Claude.
Moreover, recently Grok has been under fire for creating sexualized images, demonstrating that the model could easily be exploited by bad actors and lead to “data poisoning.”
Industry experts and think tanks also suggest that Grok may not yet be a “peer” to Claude in terms of performance and reliability for complex Department of Defense tasks.
According to Gregory Allen, a senior adviser focused on AI at the Center for Strategic and International Studies think tank, “I do not believe they are peers in performance right now across all of the capabilities that matter to a customer like the Department of War.”
Despite the xAI logo appearing on the GSA’s experimental platform for federal employees, Grok remains excluded due to safety concerns.