According to 1M AI News monitoring, before OpenAI officially lifted the ban on military use in January 2024, the Pentagon was already using OpenAI models through Microsoft Azure. The Pentagon did not need to negotiate directly with OpenAI and could access GPT series models using existing Azure cloud service contracts. Microsoft has independent licensing terms for OpenAI products on Azure, and both companies’ spokespeople confirmed that Azure OpenAI products are not, and have never been, subject to OpenAI’s own usage policies.
Since 2019, Microsoft has invested a total of $1.3 billion in OpenAI and distributes its models to enterprise customers via Azure. Some OpenAI employees previously discovered that the Pentagon was testing models through Azure, but most employees were unaware at the time. In January 2024, OpenAI announced support for certain government defense projects, effectively formalizing the company’s policy to align with the Pentagon’s existing usage practices.