BLM (Bureau of Land Management), which is federal land managed by the Department of the Interior. This is about 10% of US land.

Why I am saying it as theater is that they were one of the firsts to deploy their models on government classified networks, at National Laboratories.

The discussion is if that they want to be able to veto actions that the US government take if they don't like it. That is basically the crux of the matter and what the disagreement is about.

Anthropic does not disclose, for security and legals reason, on what data they train their models on. AI companies talk a good game about privacy, but their whole business model is based on getting more data data, including private data, to improve their models.

I am also just like you a foreigner looking in. Since living in the EU I have developed a deeper distrust of unelected bureaucrats, than for politicians that can at least be removed.