As we mentioned yesterday, Senator Elizabeth Warren is rightly outraged at the Pentagon’s recent plan. Indeed, they’ve gone so far as to let Elon Musk’s new company, xAI, on to classified networks. In the letter, which was addressed to Defense Secretary Pete Hegseth, Warren warned about the national security risks associated with the artificial intelligence model Grok. This announcement follows the filing of a class action lawsuit against xAI on the very same day. The US lawsuit claims that Grok produced commercial sexual content by prompting the service with actual images of the plaintiffs when they were minors.
Warren’s letter urges the Department of Defense (DoD) to be more transparent. He’s looking for information on the agreement entered into with xAI to use Grok within classified systems. She is seeking more detailed information on the department’s plans to protect Grok from cyber attacks. She asks them how they plan to stop it from leaking sensitive military information.
During the hearing, the senator did not mince words about her objections. This would not only put Grok’s safety guardrails at great risk, but pose grave threats to the safety of U.S. military personnel and the cybersecurity of classified systems. Her statement further emphasizes the need to make sure that AI technologies used in defense environments are carefully tested.
A coalition of nonprofits has echoed Warren’s concerns, urging the government to halt the deployment of Grok within federal agencies, including the DoD. This coalition points to alarming instances where users have prompted the chatbot to create sexualized images of women and children without their consent.
Adding to the controversy, reports surfaced of a former employee from Musk’s Department of Government Efficiency allegedly stealing personal data from the Social Security Administration, further complicating the narrative surrounding xAI and its operations.
In her initial letter, Warren asked what guarantees that xAI has provided to the DoD. She wanted to test Grok’s security measures and data handling procedures, in particular. She questioned whether these assurances were adequately assessed prior to allowing access to classified networks.
The Pentagon has repeatedly expressed strong confidence in Grok’s capabilities. Chief Pentagon spokesperson Sean Parnell stated that the DoD “looks forward to deploying Grok to its official AI platform GenAI.mil in the very near future.”
Rebecca Bellan, a senior reporter at TechCrunch, has been all over the news coming out about Grok and what it means. Her investigative reports have pioneered attention on the business and policy interests that are actively shaping artificial intelligence today.
If you have any questions, email all tips and press inquiries to rebecca.bellan@techcrunch.com. Or, if you prefer, you can contact her via encrypted Signal message at rebeccabellan.491.