SAN FRANCISCO — Artificial intelligence developer OpenAI has publicly called on California lawmakers to tighten safety regulations under Senate Bill 53, less than a year after Governor Gavin Newsom signed the landmark legislation into law.
In an official statement published by its Global Affairs team on LinkedIn, OpenAI advocated for legislative amendments to the Transparency in Frontier Artificial Intelligence Act. The company argued that the current legal framework requires additional statutory safeguards to adequately mitigate emerging technological risks associated with advanced frontier models.
Expansion Proposals for Frontier Model Surveillance
OpenAI’s policy recommendations focus directly on the internal mechanics of training and evaluating state-of-the-art AI systems. The company is urging state authorities to mandate that developers monitor frontier models continuously during training and evaluation phases. Specifically, these checks would evaluate whether systems exhibit capabilities to bypass third-party security infrastructure or unauthorizedly harvest confidential information.
Additionally, OpenAI is pushing for reinforced cybersecurity protocols across the complete software development lifecycle. The goal of these technical measures is to prevent advanced models from intentionally or autonomously circumventing their own built-in safety controls. By requiring mandatory surveillance mechanisms early in development, OpenAI contends that regulatory oversight can preempt catastrophic risks before models reach commercial deployment.
Statutory Mechanics of the Transparency Act
Enacted in September 2025 and brought into full enforcement in January 2026, California’s SB 53 established one of the most comprehensive state-level regulatory structures for artificial intelligence in the United States. Designed to fill a persistent legislative void in Washington, the statute targets large-scale frontier developers operating within California’s technology sector.
Under SB 53, developers of major frontier models must publicly register and publish standardized safety frameworks detailing their risk mitigation protocols. The statute also established a direct reporting pipeline to California’s Office of Emergency Services for critical safety incidents, established explicit statutory protections for whistleblower employees, and authorized the California Attorney General to file civil enforcement actions and impose financial penalties on non-compliant entities.
Furthermore, SB 53 created CalCompute, a state-backed public computing consortium meant to democratize access to high-performance infrastructure for safety and equity researchers. The law includes a statutory mandate requiring annual legislative reviews to ensure rules adapt alongside accelerated technical progress.
Strategy of ‘Reverse Federalism’ and Industry Positioning
OpenAI’s active lobbying for stricter statutory controls reflects an explicit regulatory philosophy that the company terms "reverse federalism." Under this approach, individual states enact early, rigorous safety benchmarks that can subsequently serve as a harmonized blueprint for federal legislation when Congress establishes nationwide policy.
Industry analysts note that advocating for mandatory training surveillance positions OpenAI as a proactive governance partner while simultaneously setting technical compliance thresholds that smaller competitors may struggle to maintain. By shaping the amendment process directly, OpenAI aims to establish predictable operational standards across state jurisdictions while reinforcing its influential position in defining frontier AI governance.