AI Safety Debate: Why Working With Money and Power Matters

A leading voice in AI safety, Richard Ngo, recently suggested that past efforts failed because power and money corrupted the field. His fix: change the community's rules. But another researcher pushing back says that is the wrong move. Instead, AI safety should embrace power and money—not from a place of greed, but from realism.

The idea is to split AI safety into three arms: intellectual, political, and capital. The intellectual arm keeps its high standards and pure research. The political and capital arms do the messy work of selling safety to lawmakers, executives, and investors. That means talking to people who already handle catastrophic risks, like nuclear watchdogs, financial regulators, and pandemic experts. They know how to turn fear into concrete rules.

Why does this matter to you? Because AI is being built at civilization-scale speed. If safety experts only talk to each other, they lose the levers that actually stop harm. But if they can influence governments and insurance markets, they might force companies to slow down and build safeguards before disaster strikes.

The author admits this is a compromise. Working with power can get messy, and regulators can be captured. But the alternative—pure academics ignoring politics—leaves us unprotected. The window to act is now, and the author argues that avoiding mass economic chaos is an easy sell if you present it right.