Regulating AI via Model Constitutions

The piece argues every major model embeds a value 'constitution' (system card/alignment rubric) and that the new order targets these documents by excluding models that encode CRT, 'transgenderism,' or similar frames. This shifts governance toward rewriting the meta‑rules that shape outputs, not just moderating outputs after the fact. — It reframes AI policy as a battle over explicit value charters that vendors must present and defend to win public contracts.