Rare sight: Elon Musk actually publicly liked Mark Zuckerberg. Normally, a lot of people these days think that when it comes to the AI threat, the stronger the model, the more it should be restricted and only made available to a select few individuals or institutions. But Meta's Mark Zuckerberg believes that cutting-edge AI models shouldn’t be limited to a few people—they should be open to all of humanity. The reasoning is that the real danger comes from concentrated power. For example, a few labs, governments, or institutions controlling superintelligence is more dangerous than the models themselves. So, the safer path is to widely distribute superintelligence to billions of people, allowing everyone to use it, creating mutual checks and balances. Currently, OpenAI, Anthropic, and Google all lean toward imposing development restrictions on advanced AI. They’re worried that if capabilities advance too quickly, humanity won’t have time to understand and control them, increasing the risk of misuse. It’s expected that in the future, people will split into two camps over this perspective, and the debate will likely go on and on.