OpenAI Advocates for Stricter Rules in California's Frontier AI Legislation

Artificial intelligence research company OpenAI is calling on California lawmakers to tighten regulations on frontier AI systems, proposing updates to state legislation enacted late last year, according to details reported by Mashable Tech.

In a post published on LinkedIn by OpenAI's Global Affairs team, the organization advocated for amendments to Senate Bill 53, officially titled the Transparency in Frontier Artificial Intelligence Act. Governor Gavin Newsom signed the measure into law in September 2025. OpenAI argued that the legislation requires stricter mandates to track advanced AI models while they are being trained and evaluated.

Specifically, OpenAI recommends mandating continuous oversight during training phases to detect whether frontier models exhibit capabilities to breach third-party security defenses or extract unauthorized confidential data. The company also urged state officials to enforce enhanced cybersecurity protocols throughout the entire development lifecycle to block AI systems from overriding their own internal safety guardrails.

The existing SB 53 statute already established significant compliance burdens for major AI developers in California. Under the law, companies developing frontier models must publicly release their safety protocols and report severe safety incidents to the California Office of Emergency Services. The legislation further safeguards whistleblowers and authorizes the state Attorney General to issue civil fines to non-compliant firms.

Additionally, SB 53 instituted CalCompute, a state-backed public computing initiative designed to foster equity and safety research in artificial intelligence. The law mandates annual reviews by state officials to adjust regulatory standards in response to technological developments. At the time of its signing, Newsom presented the measure as a balanced approach to consumer protection and economic growth in the AI sector, acting as a regional substitute for stalled federal legislation.

OpenAI described its regulatory approach as "reverse federalism," advocating for state-level standard-setting that could eventually inform unified federal regulations while Congress delays comprehensive AI legislation. The company emphasized that its proposals are proactive rather than a reaction to a specific safety event, aiming to establish early-detection protocols and promote information sharing across the artificial intelligence sector.

The policy push comes amidst ongoing legal scrutiny for the San Francisco-based company. Mashable noted in its coverage that its parent entity, Ziff Davis, initiated a copyright infringement lawsuit against OpenAI in April 2025, claiming the company impermissibly used copyrighted material to train and operate its generative AI platforms.