* Whoever controls it can leverage it to gain a competitive advantage and reinforce inequalities: a country can decide to prevent other countries from using it.
* Whoever controls it doesn't have any guardrails (by definition: they set the guardrails), and is receiving a massive amount of data from all their users. Say most of the world is using LLMs based in the US, it means that those companies have access to the data exposed by most of the world in a system that is very good at leveraging information. So the US government (in this case, but any government with control over the major players really) can leverage that information to get a competitive advantage.
What happens if you get full access to the technology of a major LLM company to the point where you can ask questions like "here is a list of people close to that political opponent, please give me a list of actions that can be taken to ruin their career, from the most legal to the least legal"?
Who says that Huawei is a national security risk? Seems to me that LLMs are orders of magnitude more dangerous.