I’ve been watching ($OPG ) with a lot of curiosity lately, not because it feels completely new, but because it’s trying to formalize a problem I’ve already seen quietly exist in today’s AI systems. I keep coming back to their core idea that AI isn’t really “owned” by users—it’s something we’re granted access to, and that access can be limited, changed, or removed depending on whoever controls the infrastructure. That observation feels real to me, even if it’s uncomfortable.
What interests me more is how they try to respond to it. Using things like TEEs and zkML sounds technical on the surface, but at a deeper level it’s really about trust—how I can trust an AI system without fully seeing what happens inside it. TEEs ask me to trust hardware boundaries, while zkML tries to replace trust with proof. I see the logic, but I also know these layers rarely stay simple once they meet real-world pressure.
When I hear “censorship-resistant AI,” I don’t think of slogans. I think about all the hidden places where control can still exist—compute, updates, access, economics. So for me, $OPG feels less like a finished idea and more like an ongoing attempt to answer a very hard question: can AI ever be truly open in practice, not just in theory?
