Congress May Only Get One Shot at AI Rules

Hundreds of bills about artificial intelligence — computer systems that can write, draw and make decisions — have been introduced in Congress. Almost all die in committee without a vote. Most tackle just one slice of the problem: data centers, children's safety, or making companies more transparent about how AI works. The author points to the fight over releasing the Epstein Files as proof of how this goes: even massive public outrage produced limited results, and momentum faded.

So he argues we should assume we get one real shot at passing something substantial about AI. His draft law would create temporary expert panels covering 19 different AI topics — from today's everyday annoyances to long-term, even existential risks, at home and abroad. Each panel must deliver findings by a deadline. Then, using "hammer provisions" — existing congressional rules that force a vote once certain conditions are met — Congress would be required to actually act on that advice instead of shelving it.

After that, the panels would become a permanent AI Council plus an international diplomatic body. Their job: keep evidence on record, oversee what's happening, and prepare future laws in advance so they're ready when needed. The author is blunt about why this matters: he fears AI power concentrating in a few hands could help create an authoritarian state. He also admits his draft is incomplete and asks others working on similar ideas to team up.

He also pushes back on the popular rallying cry to "pause" AI development. His worry is simple: if a pause passes, public attention and political will will cool off, and the chance to do more disappears. He wants a complete package ready right now, while people are still angry and paying attention.