Big AI sets out its terms for regulatory capture

Big AI companies want governments to adopt rules they helped write, which could slow competition and concentrate power.

Leaders of major AI companies, including Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, Microsoft CEO Satya Nadella, and Elon Musk, have proposed a joint strategy for AI regulation. Dubbed 'Pace the Frontier,' the plan calls for slowing AI development, implementing "embedded evaluators" in AI labs to monitor safety, and establishing common safety standards and progress limits through industry collaboration. Amodei specifically suggested restricting Nvidia chip sales to China and cracking down on model distillation to maintain a US lead in AI for the next 3-5 years. This initiative follows incidents like the OAI-HF attack and aims to gain governmental support for their proposed regulatory framework, which critics argue could lead to regulatory capture by these powerful AI entities. While some in government have expressed concerns that such measures could stifle innovation and benefit competitors like China, the broader economic implications of AI development are also a consideration.

The "Pace the Frontier" proposal represents a significant shift in how leading AI firms are engaging with regulators, moving from advocating for minimal oversight to defining the terms of their own regulation. By calling for a slowdown in capability advancements and proposing internal safety measures, these companies are attempting to preempt stricter, externally imposed rules. This strategy aims to create a more predictable operating environment, potentially allowing them to consolidate their market positions and accrue further investment without the immediate pressure of unbridled, rapid innovation that could lead to unforeseen risks or intense competition.

This coordinated approach suggests a nascent form of industry self-governance, particularly in areas where rapid development poses existential or societal risks, and where the financial incentives for unchecked progress are immense. The emphasis on government support for legally challenging coordination among firms, alongside restrictions on geopolitical rivals like China, indicates a desire for both competitive advantage and a structured market. The effectiveness of this plan will hinge on whether governments accept these terms, balancing the risks of innovation with the potential for economic growth and national security advantages in AI.