OpenAI Urges Federal AI Standards, Citing Risk of 'Patchwork' State Regulations

Miles Brundage, a prominent voice in the AI policy sphere, recently criticized OpenAI's communication with California Governor Gavin Newsom regarding Senate Bill 53 (SB 53), stating on social media that the letter was "filled with misleading garbage about SB 53 and AI policy generally." This public rebuke highlights the ongoing tension between AI developers and state regulators over the future of artificial intelligence governance. Brundage's tweet underscored a sentiment among some observers that the company's actions might be "unsurprising if you follow this stuff."

California's SB 53, known as the Transparency in Frontier Artificial Intelligence Act, aims to establish a framework for AI safety and accountability within the state. The bill mandates transparency requirements for large AI developers, including the publication of safety and security protocols, and requires reporting of critical safety incidents to the Attorney General. It also introduces whistleblower protections for employees who disclose significant risks and proposes the creation of "CalCompute," a public cloud computing cluster to democratize access to AI tools for startups and researchers.

OpenAI, a leading AI company, has consistently advocated for a harmonized federal approach to AI regulation, expressing concerns that a "patchwork of state rules" could stifle innovation. In their letter to Governor Newsom, OpenAI argued that state-by-state regulations might impede technological advancement and hinder the United States' competitive edge. The company suggested that compliance with federal and global safety guidelines should suffice for developers, recommending that California treat frontier model developers as compliant if they adhere to relevant U.S. federal agency agreements or international frameworks like the EU's AI Code of Practice.

The core of the disagreement, as implied by Brundage's tweet and OpenAI's public statements, centers on whether AI regulation should be primarily a state or federal responsibility. While proponents of SB 53, including Senator Scott Wiener, argue the bill strikes a balance between fostering innovation and establishing necessary guardrails, OpenAI maintains that inconsistent state requirements could create undue compliance burdens, especially for smaller developers. This debate underscores the complex challenge of regulating a rapidly evolving technology while fostering its immense potential.