Warren presses Pentagon over decision to grant xAI access to classified networks

Sen. Elizabeth Warren (D-MA) sent a letter to Defense Secretary Pete Hegseth connected Monday expressing interest implicit the Pentagon’s reported determination to springiness Elon Musk’s institution xAI entree to classified networks.

“Grok, the arguable AI exemplary developed by xAI, has provided disturbing outputs for users, including giving users ‘advice connected however to perpetrate murders and violent attacks,’ generating antisemitic content, and creating kid intersexual maltreatment material,” the missive reads. 

Warren said Grok’s “apparent deficiency of capable guardrails” could airs “serious risks to the information of U.S. subject unit and to the cybersecurity of classified systems.” She demanded Hegseth supply accusation connected however the Department of Defense plans to “mitigate these imaginable nationalist information risks.”

Warren isn’t the archetypal to explicit alarm astatine Grok, xAI’s arguable chatbot, gaining entree to classified systems. Last month, a coalition of nonprofits urged the authorities to instantly suspend the deployment of Grok successful federal agencies, including the DoD, aft X users repeatedly prompted the chatbot to crook existent photos of women, and successful immoderate cases children, into sexualized images without their consent. The aforesaid time Warren sent her letter, a people enactment suit was filed against xAI alleging Grok had generated intersexual contented from existent images of the plaintiffs arsenic minors. 

The missive comes successful the aftermath of the Pentagon’s determination to label Anthropic a proviso concatenation risk aft the AI steadfast refused to springiness the subject unrestricted entree to its AI systems. Anthropic had been, until recently, the lone AI institution with classified-ready systems. In the midst of that conflict, the DoD signed an agreement with OpenAI arsenic good arsenic xAI to usage the 2 companies’ AI systems successful classified networks, according to Axios. 

A elder Pentagon authoritative confirmed that Grok was onboarded to beryllium utilized successful a classified setting, but is not yet being used.

“It is unclear what assurances oregon documentation xAI has provided to the Department of Defense astir Grok’s information safeguards, data-handling practices, oregon information controls, and whether DoD has evaluated those assurances earlier reportedly allowing Grok entree to classified system,” Warren writes.

Warren requested a transcript of the woody reportedly reached betwixt the DoD and xAI connected the usage of Grok successful classified systems and an mentation of however the section plans to guarantee Grok is not exposed to cyberattacks and volition “not leak delicate oregon classified subject information.”

(Last week, a erstwhile worker of Musk’s Department of Government Efficiency reportedly stole Americans’ idiosyncratic information from the Social Security Administration and stored it connected a thumbdrive — the latest accusation of DOGE-related information leakage.)

Chief Pentagon spokesperson Sean Parnell said the section “looks guardant to deploying Grok to its authoritative AI level GenAI.mil successful the precise adjacent future.”

GenAI.mil is the military’s unafraid endeavor level for generative AI that gives DoD workers entree to ample connection models (LLMs) and different AI tools wrong government-approved unreality environments. It is designed to assistance with chiefly non-classified tasks similar research, papers drafting, and information analysis.