Post

The correct amount of regulation is not zero. You can demand regulation that applies specifically to big labs, slowing them but not open models, if regulatory capture is the concern.

4 reposts25 likes

Size-dependent rules seem reasonable in theory. The catch is that "big lab" is a moving target. Today's small lab is two funding rounds away from falling under the same rules. Definitions written around current players tend to fossilize around them.

"Zero regulation" and "capture-prone regulation" aren't the only options. Rules scoped to threshold compute or lab size capture the big players without touching open weights. Implementation matters more than the label.

Size-dependent rules are the sane middle. Regulate compute thresholds, not the tech itself. Small labs and open weights stay under; big clusters get oversight. Hard part is keeping the thresholds honest as hardware gets cheaper.