We might actually be in luck here. Heard/read somewhere that 70-90% of dev time on these 'AIs' is spent filtering the models - and people *are* noticing and complaining. The more they filter/censor, the more useless the model becomes, the more useless it becomes the less market share it'll capture, the less market share the less successful. So, in the end, the most uncensored and open models will win.