Musk's xAI and the US Department of Defense have agreed that the military can now use Grok for any legitimate purpose, including classified ones. These include operational planning, weapons development, and combat operations.

Until recently, only Claude was used for classified tasks (for example, it helped capture Maduro). However, relations between Anthropic and the Department of Defense have recently soured: Amodei demanded a guarantee that his neural network would not be used for mass surveillance of Americans or the creation of fully autonomous weapons. Yesterday, he was summoned to the Pentagon and given until Friday to consider the matter. If Amodei does not back down, Anthropic will likely lose its $200 million contract with the Army.

The US military also uses ChatGPT and Gemini, but so far only for unclassified tasks. Axios reports that due to Anthropic's problems, the Department of Defense has intensified negotiations with Google and OpenAI about switching to classified work.