He asserted that relying solely on open-weight releases does not solve this imbalance and that formal regulation is necessary to manage risks.
This perspective challenges the common Silicon Valley view that regulation inevitably leads to regulatory capture, a situation where rules are designed to protect dominant companies from smaller competitors.
Amodei argued that well-crafted institutional processes can actually decentralize power by holding the largest "frontier" companies to higher standards than smaller challengers.
By implementing rigorous pre-deployment testing for the most advanced models while exempting smaller firms based on revenue or training costs, he suggests that oversight can mitigate cybersecurity and biological risks without stifling competition.
Looking ahead, Amodei expressed support for emerging federal approaches, including reported plans for pre-deployment testing of both proprietary and open-weight models as they approach the performance frontier.
He also backed the creation of an independent oversight body, similar to a financial industry self-regulatory organization, to establish "rules of the road" for the sector.
These mechanisms aim to modulate the development pace of the most powerful AI systems, potentially creating a market environment that advantages companies attempting to catch up to the industry leaders.