Mikkel Christensen
The U.S. Department of Defense is set to change course in its use of artificial intelligence in its most sensitive systems.
The U.S. Department of Defense is set to change course in its use of artificial intelligence in its most sensitive systems.
A new agreement could influence which technology companies will supply AI for classified military work in the future.
It concerns Elon Musk’s AI company xAI.
According to Axios, xAI has signed an agreement allowing the Grok model to access classified systems within the U.S. military.
A Defense Department official confirmed the deal to the outlet, while The New York Times was first to report that the agreement had been finalized.
Until now, Anthropic’s model Claude has been the only AI solution approved for systems handling sensitive intelligence, weapons development and operational planning.
Pressure on supplier
The agreement comes amid a dispute between the Pentagon and Anthropic. According to Axios, the Defense Department has required that Claude be available for “all lawful purposes.”
Anthropic has rejected the demand and has particularly sought to block use related to mass surveillance of Americans and the development of fully autonomous weapons.
xAI has accepted the requirement for “all lawful use,” Axios previously reported. The company has not publicly commented on the agreement.
Defense sources acknowledge to Axios that it would be technically difficult to replace Claude entirely. The model has been used during the so-called Maduro raid through Anthropic’s partnership with Palantir.
Tense meeting ahead
According to Axios, Defense Secretary Pete Hegseth is set to meet with Anthropic CEO Dario Amodei. Sources expect a tense discussion.
An official told the outlet that Hegseth will effectively issue an ultimatum. The Pentagon is threatening to label Anthropic a “supply chain risk” and is considering further consequences if the company does not lift its restrictions.
At the same time, the department is working to expand cooperation with other AI companies.
More players in the race
Grok, Google’s Gemini and OpenAI’s ChatGPT are already used in the military’s unclassified systems. According to both Axios and The New York Times, Google and OpenAI are negotiating access to classified systems.
The New York Times reported that Google was “close” to a deal, while OpenAI was “not close.” An official denied to Axios that Google was significantly further ahead and emphasized that discussions continue with both companies.
Sources: Axios
