Elon Musk’s AI company, xAI, has reportedly signed a deal with the Pentagon to allow the US military to use its AI model, Grok, for highly classified systems. The move comes at a moment of ‘extreme tension’ between the Department of War and Anthropic, the maker of the “Claude” AI. Reports have claimed that for months, Claude has been the only model trusted to handle the military’s most sensitive work, including weapons development and battlefield operations but a dispute over ethics and safety has left the Pentagon looking for a replacement.Citing a Defense official, Axios reported that xAI agreed to “all lawful use” standard, however, it is not clear whether xAI will be able to fully replace Anthropic, or how long that process would take. According to a report by The New York Times, Defense Secretary Pete Hegseth is set to meet Anthropic CEO Dario Amodei for a ‘a tense meeting’ at the Pentagon on Tuesday (February 24).

Why Pentagon is ‘angry’ with Anthropic

The ‘choke point’ in negotiations is a specific legal standard. The Pentagon is demanding that AI companies allow their technology to be used for “all lawful purposes”, Anthropic has refused saying that its AI model is blocked from being used for mass surveillance of American citizens and fully autonomous weapons (AI that can decide to kill without human intervention).The report said that Elon Musk’s xAI, however, has agreed to the Pentagon’s terms, positioning itself as the military’s primary alternative to Anthropic.While the Pentagon is eager to move away from Anthropic, experts say that it will be difficult because the AI is already deeply integrated into military systems as it was used during the recent raid in Venezuela to capture President Nicolás Maduro through a partnership with Palantir.

OpenAI, Google may be in the race to replace Claude

The report by The New York Times also claimed that to fill the ‘Claude’ gap, the Pentagon is accelerating talks with other tech giants including Google and OpenAI. Google is said to be “close” to a deal for its Gemini model to be used in classified systems. When it comes to OpenAI, talks have been “reignited with a new sense of urgency”, though a deal for ChatGPT is far off due to complex safety concerns.Administration officials have insisted that both Google and OpenAI will have to agree to the “all lawful purposes” criteria.