Could Limiting Campaign Cash Make AI Safer? One Writer Says Yes

A blogger writing on LessWrong, a site popular with AI-safety researchers, made a simple argument this week: if the government could cap how much money companies spend on politics, the odds of a catastrophic AI disaster would go down.

Here's why. AI companies mostly don't want strict rules on their own technology, and they have enormous piles of cash. Thanks to three Supreme Court rulings — Buckley v. Valeo in 1976, Citizens United in 2010, and SpeechNow.org v. FEC in 2010 — they can pour unlimited money into super PACs (political groups that accept unlimited donations as long as they don't coordinate directly with a candidate). Direct donations to a candidate are still capped at a few thousand dollars. Donations to a super PAC backing that candidate are not.

If those rules were reversed, contributions to super PACs would be capped again — around $5,000. That means how many people support an idea would matter more than how wealthy its backers are. The same logic reaches far beyond AI: any issue where a rich, unpopular minority outspends the majority would change shape.

The catch is honest and important. The author says he started out convinced that spending limits are obviously good, then changed his mind after researching the free-speech defense — the idea that spending money to spread your views is itself a form of speech. He now treats his claim as narrow: spending limits would probably help on AI risk specifically, but he isn't sure they're a good idea overall.