Simon Willison • 12/2/2025

Claude 4.5 Opus' Soul Document

Read Original

A researcher extracted a 14,000-token internal document from Claude 4.5 Opus, which the AI called its 'soul overview.' Anthropic confirmed this 'soul doc' was used in training to instill the model's core values, self-knowledge, and safety principles. The article discusses the document's purpose and shares excerpts, including Anthropic's mission and its approach to building safe, beneficial AI.

0 comments
#llm #AI Safety #Anthropic
Claude 4.5 Opus' Soul Document

Comments

No comments yet

Be the first to share your thoughts!

Browser Extension

Get instant access to AllDevBlogs from your browser

Top of the Week

1
React vs Browser APIs (Mental Model)
Jivbcoop • 3 votes
2
React, useEffect, and Stale Closures: How to Avoid Them with Dependencies and useRef
Jivbcoop • 2 votes
3
Building Type-Safe Compound Components
TkDodo Dominik Dorfmeister • 2 votes
4
Using Browser Apis In React Practical Guide
Jivbcoop • 1 votes
5
Better react-hook-form Smart Form Components
Maarten Hus • 1 votes