He's very clearly stating that trusting AI to a few hands was an old, naiive idea that they have evolved from. Which establishes their need to keep evolving as the technology matures.

There is a lot to criticize about OpenAI and Sama, but this isn't it.

To the benefit of OpenAI. I think LLMs would still exist, but we wouldn't have access to them.

Whether they are a net positive or a net negative is arguable. If it's a net negative, then unleashing them to the masses was maybe the danger itself.