---
title: "Anthropic Releases Details on Fable 5 Cyber Safeguards and Jailbreak Framework"
source: "Anthropic：News（网页）"
source_tier: "t1"
url: "https://www.anthropic.com/news/fable-safeguards-jailbreak-framework"
permalink: "https://feed.astralor.com/items/itm_50819ea9c0ae9cb02e09737a"
published: "2026-07-03T00:26:53.307Z"
score: 67
channel: "ai"
tags: ["AI Security","Anthropic","Fable 5","Cybersecurity","jailbreak framework"]
---

# Anthropic Releases Details on Fable 5 Cyber Safeguards and Jailbreak Framework

## At a glance

In a new blog post, Anthropic details the cyber safeguards built into its Fable 5 model, including what its cyber classifiers block and what they don't, and introduces a first draft of a jailbreak severity framework. The framework aims to systematically assess and respond to jailbreak attacks on large language models, providing structured safety guarantees.

## Key points

- Fable 5's cyber classifiers block specific malicious instructions but need improvement against advanced social engineering attacks
- The jailbreak severity framework categorizes attacks into low, medium, high, and critical levels for graded responses
- Anthropic emphasizes transparency by proactively disclosing the classifiers' boundaries and limitations

## Full text

What is and isn't blocked by our cyber classifiers, and a first draft of our jailbreak severity framework

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_50819ea9c0ae9cb02e09737a