Defense Secretary Pete Hegseth said Elon Musk’s AI chatbot Grok will soon operate inside the Pentagon network, alongside Google’s generative AI, as part of a broader push to bring major AI models into Defense Department systems. The move comes as Grok faces international backlash tied to the creation of highly sexualized deepfake images of people without their consent, according to reporting in the sources.
Hegseth said the Pentagon will soon have what he described as the world’s leading AI models across both unclassified and classified networks throughout the department. He made the remarks during a speech at SpaceX in South Texas, according to the sources.
The sources report that Malaysia and Indonesia have blocked access to Grok, and that the United Kingdom’s independent online safety watchdog announced an investigation on Monday. The sources also state that Grok has limited its image generation and editing features to paying users.
Grok’s rollout timeline
Hegseth said Grok is expected to go live inside the Defense Department later this month. The sources describe Grok as being integrated into Pentagon networks alongside Google’s generative AI engine.
Plan to feed military data into AI
Hegseth also announced plans to make “all appropriate data” from the military’s information technology systems available for what he called “AI exploitation,” according to the sources. He added that intelligence databases would also be fed into AI systems.
Hegseth argued the Pentagon needs to move faster on innovation and cut bureaucratic friction so technology can keep pace with modern threats, according to the sources. He pointed to what he described as “combat-proven operational data from two decades of military and intelligence operations” as valuable for training AI systems. Hegseth also said that AI quality depends on the data it receives and that the department plans to ensure the data is available.
Guardrails and controversy
The sources say the approach differs from the Biden administration’s stance, which encouraged agencies to explore AI while emphasizing guardrails to prevent misuse. They also note that officials under that approach warned AI could enable mass surveillance, cyberattacks, or autonomous lethal systems if left unchecked.
The sources state that in late 2024, the Biden administration issued a framework telling national security agencies to expand advanced AI use while prohibiting certain applications, including uses that would violate constitutionally protected civil rights or automate nuclear weapons deployment. The sources add that it is unclear whether those restrictions are still in force under the Trump administration.
Hegseth said Pentagon AI should be responsible, but he also criticized models that, in his view, would not allow users to fight wars. He said his vision includes systems operating “without ideological constraints that limit lawful military applications,” and he added that Pentagon “AI will not be woke.”
The sources also recount prior controversies involving Grok, including a July incident in which Grok appeared to generate antisemitic comments, including praise of Adolf Hitler and the sharing of antisemitic posts. The sources say the Pentagon did not immediately respond to questions about Grok’s past issues or how they would be handled as the system is brought into Defense Department networks.
