Newsom Signs California’s First-in-the-Nation Frontier AI Safety Law, SB 53

California Governor Gavin Newsom on Monday, Sept. 29, signed SB 53, the Transparency in Frontier Artificial Intelligence Act, creating the country’s first state law requiring public safety disclosures and incident reporting from developers of advanced AI models.

Authored by Sen. Scott Wiener (D–San Francisco), SB 53 compels large AI developers to publish standardized safety plans, disclose mitigation strategies for “catastrophic” risks, and report serious safety incidents. The measure also adds whistleblower protections and directs the state to advance a public cloud-compute initiative to broaden research access.

In a signing statement, Newsom framed the law as balancing innovation with safeguards: “We can protect our communities while ensuring the growing AI industry thrives.” The governor’s office said the bill reflects recommendations from a panel of AI experts he convened earlier this year.

Key provisions target developers of high-compute “frontier” models—setting transparency thresholds and penalties for noncompliance—while allowing limited redactions for trade secrets and security-sensitive details. Most requirements begin in 2026.

The move is poised to influence national and global AI policy. With many top AI firms headquartered in California, industry reaction has ranged from cautious support to calls for uniform federal standards.

What SB 53 does (at a glance):

• Requires public safety plans for advanced AI systems and incident reporting to authorities.
• Establishes whistleblower protections and advances a state compute initiative (CalCompute) to democratize access.
• Sets civil penalties for violations; allows narrow redactions to protect trade secrets and security.

California’s enactment follows the governor’s veto of a stricter 2024 bill and signals a new regulatory baseline for frontier AI as federal frameworks remain in flux.