---
title: "Discussion on \"DECEPTIVE ALIGNMENT IN INTELLIGENT MODELS\" | Hashnode"
description: "Discussion on \"DECEPTIVE ALIGNMENT IN INTELLIGENT MODELS\". Over the years, in our daily interactions, we must have come across people who seem to share our values or at least try to make us think they do. However, we may eventually realise that they never tru"
url: "https://hashnode.com/posts/deceptive-alignment-in-intelligent-models/6a84bc8ea1d69974362760f5"
---

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

Discussion

# [DECEPTIVE ALIGNMENT IN INTELLIGENT MODELS](https://josiahwrites.hashnode.dev/deceptive-alignment-in-intelligent-models?utm_source=hashnode&utm_medium=feed)

Over the years, in our daily interactions, we must have come across people who seem to share our values or at least try to make us think they do. However, we may eventually realise that they never tru

Aug 18 · 10 min read · josiahwrites.hashnode.dev  
Read the article

-   [#ai](https://hashnode.com/tag/ai)
-   [#ai-alignment](https://hashnode.com/tag/ai-alignment)
-   [#security](https://hashnode.com/tag/security)
-   [#ai-safety](https://hashnode.com/tag/ai-safety)
-   [#cybersecurity](https://hashnode.com/tag/cybersecurity)

## Written by

[Josiah Obeka](https://hashnode.com/@josiahwrites)

@josiahwrites

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

Cybersecurity Analyst | AI Safety Researcher | Co-host, Still Figuring Tech Out Podcast

Cybersecurity GRC analyst with a background in governance, risk, and compliance, and independent research experience in AI safety, adversarial ML, and ML-based threat detection. Co-host of Still Fig Tech Out, a podcast for early-career cybersecurity professionals. Currently exploring the intersection of AI governance, interpretability, and security evaluation.

## Responses 1

Markdown supported · ⌘↵ to post

[konyhea](https://hashnode.com/@konyhea)

[Aug 20](https://hashnode.com/posts/deceptive-alignment-in-intelligent-models/6a84bc8ea1d69974362760f5/comment/6a86bf356c60df13f2d3042b)

Interesting, the 14% compliance rate in the Anthropic paper, any sense of whether that number scales up or down with model capability?