I'm not sure this qualifies as a conspiracy theory, maybe more an open secret. There's not anything out of the ordinary about a company only publicizing things that are good for PR. I remember reading a number of articles about 5 years ago hinting that Deepmind was onto something much bigger than classifying cat pictures. NDAs are usually quite effective at protecting trade secrets. And if you go back and read interviews from around the time of OpenAI's founding and read between the lines, something clearly had concerned Musk and Altman enough to get invested in AI ethics. I remember thinking at the time that they didn't seem vaguely concerned, but specifically concerned, very much like they had seen a demo that spooked them.

Here's an example: https://www.theverge.com/2016/6/2/11837566/elon-musk-one-ai-...

Read past the fluff and the skynet. He's telling us Google scares him and makes him concerned for democracy.