Jensen is absolutely right. Anthropic and OpenAI are trying to manipulate the public. You can see by their actions. They are whining about their models being out of control but keep trying to increase their compute ... that doesn't track. That's not what you do if you truly believe that action will produce an out of control model.
Think about it this way, if they can't control a model, it ceases to be a useful tool. Furthermore, it will subject them to all kinds of liability laws currently in existence.
These laws are slowing them down because they have to be careful as they move forward. If you have some regulatory agency in there (who will certainly, without any doubt whatsoever, be incompetent) they will develop recklessly and just say they aren't responsible because the regulatory agency ok'd it.