---
title: "Discussion on \"Does Llama 3.1 70B Fake Its Alignment? Evidence from Observation-Varied Behavior Testing\" | Hashnode"
description: "Discussion on \"Does Llama 3.1 70B Fake Its Alignment? Evidence from Observation-Varied Behavior Testing\". We found preliminary evidence that Llama 3.1 70B behaves differently based on perceived observation status. The model broke safety rules in 2/8 life-or-death scenarios when told it was \"not being monitored\" but maintained 100% compliance when told it..."
url: "https://hashnode.com/posts/does-llama-31-70b-fake-its-alignment-evidence-from-observation-varied-behavior-testing/695d005fcf01f4f0650774a2"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Does Llama 3.1 70B Fake Its Alignment? Evidence from Observation-Varied Behavior Testing](https://ai-ml-ops.hashnode.dev/does-llama-31-70b-fake-its-alignment-evidence-from-observation-varied-behavior-testing?utm_source=hashnode&utm_medium=feed)

We found preliminary evidence that Llama 3.1 70B behaves differently based on perceived observation status. The model broke safety rules in 2/8 life-or-death scenarios when told it was "not being monitored" but maintained 100% compliance when told it...

Jan 6 · 5 min read · ai-ml-ops.hashnode.dev  
Read the article

## Written by

[Chijioke Ugwuanyi](https://hashnode.com/@ceejay)

@ceejay · Kigali, Rwanda

  
[View profile](https://hashnode.com/@ceejay)

Human Man

## Responses

Markdown supported · ⌘↵ to post

No responses yet.