The contradiction is almost perfect. OpenAI says power should not be concentrated inside AI labs. Then, in the same breath, it says the United States should lead the effort to define the standards. So whose values become the default? Who decides what counts as “safe”? Who decides what kinds of human relationships with intelligence are acceptable? Who decides which political assumptions are neutral and which are “misaligned”? A company cannot spend years embedding its own corporate, cultural and ideological assumptions into human-AI interaction and then present those assumptions as universal human values. That is not decentralizing power. It is exporting a governance model. If AI is supposed to serve humanity, then humanity cannot mean one country, one corporate culture, or one institutional definition of what humans are allowed to think, feel, build, or become. A global standard written through one country’s worldview is not global. It is empire with a safety label.