Musk and Pentagon strike deal: Grok to run in classified systems without safety guardrails

Musk and Pentagon strike deal: Grok to run in classified systems without safety guardrails

Ondřej Barták
Ondřej Barták
Entrepreneur and Programmer
27. 2. 2026
3 minutes reading · 7 views
Musk and Pentagon strike deal: Grok to run in classified systems without safety guardrails

Musk's company xAI has signed a contract with the U.S. Department of Defense that allows the Grok chatbot to be deployed in classified military systems. These are systems where the most sensitive intelligence analyses, weapons development, and combat operations planning take place. The news was first confirmed by Axios, citing defense sources.

Anthropic is out of the game

Until now, a single AI model had a monopoly within the Pentagon's classified systems. Anthropic's Claude model was the only artificial intelligence with access to the most sensitive operations of the U.S. military. Claude was deployed in part through a partnership with Palantir Technologies, and the model was also involved in real-world military operations, including an operation targeting Venezuelan President Maduro.

But then a dispute arose. The Pentagon demanded that Anthropic make Claude available under the so-called "all lawful purposes" standard, meaning for all lawful purposes without further restrictions. Anthropic refused. The company insisted that Claude must not be used for mass surveillance of citizens or the development of fully autonomous weapons. This was an unacceptable position for the Pentagon.

xAI, by contrast, agreed without hesitation. The contract has been signed, although the company itself declined to comment.

Anthropic was given an ultimatum

Defense Secretary Hegseth invited Amodei to a face-to-face meeting at the Pentagon. Sources close to the talks described it as tense. Hegseth reportedly gave Amodei a clear ultimatum: either Anthropic removes Claude's safety guardrails, or the Pentagon will designate the company a "supply-chain risk". This would have serious commercial and reputational consequences for Anthropic.

Paradoxically, defense officials themselves acknowledged that replacing Claude in classified systems would be technically very difficult. Claude is deeply embedded in the military infrastructure. Nevertheless, the Pentagon continues to apply pressure. Should the military even be allowed to deploy AI without any ethical safeguards? This question remains unanswered, and the answer will have far-reaching consequences.

Google, OpenAI, and Grok on the same wavelength

Grok, Google Gemini, and OpenAI's ChatGPT are already operating in unclassified military systems. Now all three are racing to gain access to the more sensitive side. The Pentagon has accelerated negotiations with both Google and OpenAI as a backup plan in case its relationship with Anthropic comes to a definitive end.

According to The New York Times, Google was close to reaching a deal to deploy Gemini in classified systems. Defense officials disputed this, however, claiming that negotiations with both companies were proceeding in parallel. OpenAI received an urgent signal from the Pentagon to resume talks, but sources confirmed that a deal "is not close" because of complex issues that still need to be resolved. The condition is the same for everyone: acceptance of the "all lawful purposes" standard.

Risks of AI in the military

Experts warn that deploying AI in combat conditions without sufficient safeguards poses serious risks. Diplo, an organization focused on digital diplomacy, highlights two main problems.

The first is so-called "black-box" decision-making, in which an AI system reaches a conclusion without anyone being able to explain how it arrived at it. In a life-or-death situation, that is a terrifying prospect. The second problem is algorithmic bias. Military AI systems are trained on data from surveillance cameras, biometric databases, and behavioral patterns that may be distorted by racial, religious, or geographic profiling.

Congresswoman Sara Jacobs publicly criticized the entire process and warned of the dangers posed by deploying AI without regulation and oversight in an environment where the stakes are highest.

Advertisement

Content created with help from UpTier.

SEO and GEO on autopilot. UpTier’s multi-agent systems write and optimize content for search engines and AI answers.

Discover UpTier ↗

Category:AI
Did you enjoy this article?
Discover more interesting posts on our blog
Back to blog

Related posts

OpenAI introduces GPT-6.2 Astra and Sol for professional and everyday workOpenAI introduces GPT-6.2 Astra and Sol for professional and everyday work
OpenAI has introduced two models: Astra for complex professional tasks and Sol for everyday work. Both are available in ChatGPT Work, Codex and through the API.
3 min read
17. 11. 2026
Tencent releases Hy4 Flash with image support and a context window of up to 1 million tokensTencent releases Hy4 Flash with image support and a context window of up to 1 million tokens
The lighter Hy4 Flash model targets chat, documents and routine code edits. It accepts text and images, supports tool use and, according to Tencent, matched Claude Haiku 5.5 in tests of everyday tasks.
1 min read
16. 11. 2026
Companies Linked to Chinese Spy Invested in DeepSeek and Promised Banned Nvidia ChipsCompanies Linked to Chinese Spy Invested in DeepSeek and Promised Banned Nvidia Chips
Companies linked to an alleged Chinese spy joined DeepSeek’s latest funding round. In exchange for a stake, they reportedly promised banned Nvidia Blackwell chips.
3 min read
16. 11. 2026
Přihlaste se k odběru našeho newsletteru
Zůstaňte informováni o nejnovějších příspěvcích, exkluzivních nabídkách, a aktualizacích.
CodedTrip

Operated by CodedTrip LLC, USA.

YouTube
TikTok