>The real story here is that this may be the beginning of governments restricting the availability of strong LLMs to the public, to you.

I can't agree more. This is a precedent not just in denial but possible vagueness. Judiciaries have 'vagueness doctrines' to counter such laws/directives but _these_ may be re-trumped by the deference given to national security.

If we don't get soon a framework by which models may be measured as 'too powerful' vs 'not too powerful' we supercharge the self-dealing (corruption) that this administration has brazenly adopted. Many fingers can be put on many scales; groks may be given a pass while others are held to higher "standards".

Will OpenAI now just asymptotically bump its versions to 5.99999999 to stay under a limit that nobody really understands?

I realize that this has all just happened and we might get some good rigorous clarification from our government.... sigh. We are living in a kakistocracy. Who am I kidding?