Interesting - OpenAI rewriting its preparedness framework post HF (and after one of its systems plausibly hit critical on cyber).
One element to watch is which revisions they put in their legally binding framework and which they put in their separate voluntary framework.