---
title: "Frontier Safety Framework"
description: "Google DeepMind's published policy for severe risks from its most capable models: Critical Capability Levels for misuse, machine learning R&D and misalignment, early-warning evaluations, and…"
canonical: https://aigovernanceengineer.com/glossary/frontier-safety-framework
author: "Jorge García Aibar"
license: "CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)"
doi: https://doi.org/10.5281/zenodo.22956197
version: "0.5.0"
updated: 2026-09-30
---

# Frontier Safety Framework

Google DeepMind's published policy for severe risks from its most capable models: Critical Capability Levels for misuse, machine learning R&D and misalignment, early-warning evaluations, and security and deployment mitigations, reviewed at least once a year. Version 3.1 is dated 17 Apr 2026 [1].

- Developed in: [the frontier safety frameworks crosswalk](https://aigovernanceengineer.com/resources/frontier-safety-crosswalk)
- Chapters: [ch. 10, Reading List](https://aigovernanceengineer.com/bok/reading-list)
- In the glossary chapter: https://aigovernanceengineer.com/bok/glossary#t-frontier-safety-framework

## Sources

[1] Frontier Safety Framework, Version 3.1 (Critical and Tracked Capability Levels, security and deployment mitigations, annual review). Google DeepMind. 2026-04-17. https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf (verified: primary)