Pentagon to Deploy Musk’s Grok AI on Classified Networks Despite Global Deepfake Bans
At a SpaceX facility in South Texas, Pentagon official Hegseth announced the department will deploy Elon Musk’s Grok AI chatbot — operating alongside Google’s generative AI engine — on both unclassified and classified networks and make "all appropriate data" from military IT systems and intelligence databases available for AI exploitation. The move comes amid global backlash over sexualized non‑consensual deepfakes and bans or investigations in countries including Malaysia, Indonesia and the UK, and contrasts with parts of the Biden administration’s 2024 AI framework that barred certain national‑security uses, while Hegseth framed the rollout as rejecting "woke" and "ideological constraints."
📌 Key Facts
- At a Monday event held at a SpaceX facility in South Texas, Undersecretary Hegseth announced the Pentagon will deploy Elon Musk’s Grok AI and used culture‑war language in his speech, saying DoD AI "will not be woke" and will avoid "ideological constraints."
- Grok will run alongside Google’s generative AI engine on both unclassified and classified Pentagon networks.
- Hegseth said the department will make "all appropriate data" from military IT systems and intelligence databases available for "AI exploitation," effectively wiring the systems into sensitive data sources.
- The rollout comes amid global backlash to Grok, including reports of sexualized non‑consensual deepfakes, Malaysia and Indonesia blocking the service, and a formal investigation by the U.K. regulator Ofcom.
- NPR highlighted a contrast with the Biden administration’s 2024 AI framework, which banned certain national‑security AI uses (for example, automating nuclear‑weapons deployment), and noted it is unclear whether those prohibitions remain in force under the current administration.
📊 Relevant Data
96% of deepfakes are nonconsensual sexual deepfakes, with 99-100% of victims being female.
Deepfake Statistics 2025: The Data Behind the AI Fraud Wave — DeepStrike
In 2024, there were 1,938 anti-Jewish hate crimes reported by the FBI, up 5.8% from 1,832 in 2023.
New report warns extremist groups increasingly using AI to intensify antisemitic propaganda — CBS News
AI systems in military applications introduce risks of technical failures that could lead to miscalculation and escalation, with failures interacting with strategic and human factors.
Reducing the Risks of Artificial Intelligence for Military Decision Advantage — Center for Security and Emerging Technology
Deepfake fraud incidents rose tenfold from 2022 to 2023, with 88% in the crypto sector.
2024 Deepfakes Guide and Statistics — Security.org
Men reported significantly higher rates of victimization and perpetration in deepfake creation and threats, but women are primarily targeted in sexualized deepfakes.
Non-Consensual Synthetic Intimate Imagery: Prevalence, Attitudes, and Knowledge in 10 Countries — ACM Digital Library
📊 Analysis & Commentary (1)
"A cautionary opinion arguing that hasty deployment of commercial generative AI (exemplified by plans to run Grok on classified networks) risks catastrophic outcomes because capabilities, incentives, and geopolitics have outstripped safeguards and oversight."
📰 Source Timeline (2)
Follow how coverage of this story developed over time
- Confirms Hegseth’s announcement was made Monday at a SpaceX facility in South Texas and quotes from that speech.
- Spells out that Grok will operate alongside Google’s generative AI engine on both unclassified and classified Pentagon networks.
- Adds that Hegseth will make 'all appropriate data' from military IT systems and intelligence databases available for 'AI exploitation.'
- Details the immediate context of Grok’s global backlash: sexualized non‑consensual deepfakes, Malaysia and Indonesia blocking Grok, and a formal UK Ofcom investigation.
- Explicitly contrasts Hegseth’s approach with the Biden administration’s 2024 AI framework that banned certain national-security uses (like automating nuclear‑weapons deployment) and notes it is unclear whether those prohibitions remain in force under Trump.
- Quotes Hegseth’s culture‑war framing that DoD AI 'will not be woke' and will avoid 'ideological constraints' even as it is wired into sensitive data.