Large frontier developers must publish a frontier AI framework

Practical action

Publish a safety framework and keep a change log of revisions.

Who does it apply to?

Large frontier developers above the statutory compute and revenue thresholds.

Actors

General-purpose AI model provider

Sectors

Cross-sector / all sectors

Use cases

Generative AI and foundation models

Applies from: 1 January 2026

Which controls meet this duty?

Satisfies: the control, operated properly, does the work the duty asks for. Supports: it contributes but the duty needs more. Each control page lists every other duty it serves, so work done once can be counted once.

• satisfiesPolicyHead of AI safety · annual
  Frontier model safety and security framework

  Serves 9 recorded duties · evidence: Published frontier safety framework, Dangerous-capability evaluation report, Threshold notification to an authority

  The published, annually reviewed framework with a change log.

• supportsTechnical measureAI security or safety lead · at launch and on material change
  Adversarial and red-team testing for generative AI

  Serves 8 recorded duties · evidence: Red-team exercise report, Red-team rules of engagement and scenario library, Adversarial findings tracker

  Catastrophic-risk assessment evidence.

Templates that cover this duty

Generated from the records. Free · no account · links emailed to your work address. This duty is cited in each file with its source reference and a link back here.

• AI Risk Register (MIT AI Risk Repository taxonomy)

  Register · XLSX · v4

• NIST AI RMF ↔ EU AI Act ↔ ISO/IEC 42001 Crosswalk

  Crosswalk · XLSX · v3

• AI Audit Evidence Tracker

  Register · XLSX · v1

• AI Red-Team and Evaluation Test Plan

  Assessment · XLSX and DOCX · v3

All 7 templates that cover it →

What evidence would a reviewer expect?

Framework mappings

Original editorial crosswalks. They cite clause numbers only and reproduce no standard text; confidence reflects how direct the mapping is.

See every California (United States) duty mapped this way →

Frequently asked questions

Cite this record

AIPolicyTracker (2026). “Large frontier developers must publish a frontier AI framework (California SB 53)” [AIPT-OBL-0079, verified 26 September 2026 by Bhaskar Bhatt]. https://aipolicytracker.org/id/AIPT-OBL-0079 (accessed 27 October 2026). Data licensed CC BY 4.0.

Permanent id AIPT-OBL-0079: it never changes or gets reused, and its address keeps working if this page moves.

Cite the official text alongside it: SB 53 Transparency in Frontier Artificial Intelligence Act, California Legislative Information, https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260SB53.

@misc{aipolicytracker2026_largefrontierdevelopersm,
  author       = {{AIPolicyTracker}},
  title        = {{Large frontier developers must publish a frontier AI framework (California SB 53)}},
  year         = {2026},
  howpublished = {\url{https://aipolicytracker.org/obligations/us-california-sb-53-frontier-ai-framework}},
  url          = {https://aipolicytracker.org/obligations/us-california-sb-53-frontier-ai-framework},
  urldate      = {2026-10-27},
  note         = {Data licensed CC BY 4.0. Accessed 27 October 2026}
}

Change history

Similar obligations in other instruments

• Manage systemic risk for high-impact general-purpose models — EU AI Act, European Union
• Large frontier developers must send periodic summaries of catastrophic-risk assessments to the state — California SB 53, California (United States)
• Operators of AI above the compute threshold must run lifecycle risk management and report safety results — AI Basic Act, South Korea
• Measure and test trustworthiness characteristics (Measure) — NIST AI RMF, United States (voluntary)