Post

Conversation

Ok, this may seem foolish later, but whatever, here is my conspiracy theory. I don't think OpenAI *had* to release GPT-5.6 only to trusted government partners. They said they did so "at the government's request", but what they actually did is ask them "is it ok if we release this only to you first?", the government said "sure, please do", and then OpenAI framed this as the government requesting it from them. They did this because OpenAI always tries to make themselves the center of any story about AI. They are trying to give the impression that GPT-5.6 is as capable as Mythos & Fable, and that they will thus be subject to the same regulations, and trying to make themselves part of the much bigger story about Anthropic. My theory is that they could have released GPT-5.6 in the normal way, and they chose not to, because that would confirm in the public's eye that GPT-5.6 is not on the frontier.
David Watson 🥑
Post your reply

This story doesn’t work at all. News clearly said they received calls to stop the general release
so much tin foil hat shit going on here. this government slowdown of 5.6 could also easily be framed as bearish on openai. that they couldn’t play ball with the usg any better than anthropic could and will be stuck in the same hell as them. again, just speculation
I think something like that is likely. Imagine the aura loss if they didn't have to do the same thing.
I guess we’ll see when 5.6 is released or at least a broader set of benchmarks are shown. If it is at or near mythos level generally then I’m sure they would have loved, and probably fought hard, to release it as soon as possible
if they release it and provoke an official ban, it will be way harder/take longer to get it out to everyone (Trump seems rather vindictive to me, also there would be higher political cost to retreat), if they do what USG wants, it might go easier.
this is stupid, they've said multiple times that this is not what they wanted