Van Hollen Joins Ossoff, Colleagues Launch Inquiry into Department of Defense’s Use of AI Software Known for Antisemitic Content

Today, U.S. Senator Chris Van Hollen (D-Md.) joined Senator Jon Ossoff (D-Ga.) in demanding answers as the Department of Defense plans to use AI software known for antisemitic content. Senators Van Hollen, Ossoff and their Senate colleagues launched an inquiry with Department of Defense (DoD) Secretary Pete Hegseth demanding answers over reports that the Agency plans to let the xAI chatbot, Grok, operate inside of DoD networks, despite reports that the chatbot has promoted antisemitic and other dangerous content.

Last year, there were numerous, publicly reported examples of Grok wrongfully spreading antisemitic posts and Holocaust denialism through its prompted engagement with neo-Nazi accounts.

Senators Van Hollen and Ossoff are joined by Senators Adam Schiff (D-Calif.), Richard Durbin (D-Ill.), John Hickenlooper (D-Colo.), and Reverend Raphael Warnock (D-Ga.).

"We write with concern regarding your recent announcement at SpaceX that Grok, xAI's chatbot, will join Google's generative AI engine in operating inside the Department of Defense (DoD)'s network," the group wrote to Secretary Hegseth. "We are particularly concerned by this development, given Grok's reported recent track record promoting Holocaust denial, spreading racist ideologies, and generating deepfake pornography of children."

"In the summer of 2025, Grok issued several antisemitic comments and threads, including stating that it was skeptical that six million Jews were killed in the Holocaust. More recently, the chatbot produced fake pornographic images of real people. Grok's generation of such content has triggered various investigations of the platform and the potential use of this model by a federal agency is troubling," the Senators continued.

The full text of the letter can be found here and below:

Dear Secretary Hegseth,

We write with concern regarding your recent announcement at SpaceX that Grok, xAI's chatbot, will join Google's generative AI engine in operating inside the Department of Defense (DoD)'s network. We are particularly concerned by this development, given Grok's reported recent track record promoting Holocaust denial, spreading racist ideologies, and generating deepfake pornography of children.

In the summer of 2025, Grok issued several antisemitic comments and threads, including stating that it was skeptical that six million Jews were killed in the Holocaust. More recently, the chatbot produced fake pornographic images of real people. Grok's generation of such content has triggered various investigations of the platform and the potential use of this model by a federal agency is troubling.

We therefore ask you to respond to the following questions by no later than March 2nd, 2026.

1. What guidance, processes, and restrictions have DoD placed on the use of AI within the Department? Does the Department plan to review or revise guidance with the introduction of additional AI technologies into the Pentagon network?

2. Please explain steps the Department will take to mitigate against the risk of Grok promoting antisemitic or racist conspiracy theories and tropes and sexual content?

a. Did DoD review any of these previous incidents prior to your announcement at SpaceX? Did DoD conduct a comprehensive review and comparison of Grok to other generative AI platforms? If so, what discussions and subsequent determinations did DoD have around ensuring that such incidents will not occur with DoD's use of Grok?

3. In your speech you also stated that DoD will "make all appropriate data available across federated IT systems for AI exploitation." How will DoD maintain the privacy and protect the data of servicemembers when employing Grok?

a. Are there currently any guardrails in place in regard to handling servicemember data and the use of AI technologies?

4. Will X or any third party have access to DoD prompts, outputs, logs, or corpus generated inputs used by Grok?

a. With the announcement that SpaceX has acquired xAI, how will DoD contend with SpaceX, a current defense contractor with the Department, potentially gaining access to DoD prompts, outputs, logs, or corpus generated inputs?

While AI technology can facilitate innovation, it must be deployed, used, and regulated in a manner consistent with the national interest and standards of decency. We look forward to your response and ensuring the wellbeing of the men and women in uniform who serve our nation.

Sincerely,