People have already pointed out that there's no direct correlation between "safe" and "open-source" in the other comments. I agree.

But one other reason I think this is flawed is that it assumes innovation in those models has finished. We're not at the stage where these are good enough to revolutionise everything. There's a lot more research, hard work and creativity that needs to be unleashed for all benefits to be realised. Traditionally, for-profit startups have been the best vehicle for that to happen. OpenAI has only scratched the surface and can do a lot more. Forcing them into open-sourcing and only caring about safety would quickly stop this progress. They are not a megacorp extracting rent who we need to fight - they are literally a startup changing the world in front of our eyes.