---
format: "aidr-story-markdown/v1"
id: "388ecd805eb731f877434920126141c908f5f4dee8e9d79f5261f31c5619354c"
canonical_url: "https://aidr.today/388ecd80?lang=en"
title: "Two Neuron Adjustments Turn Alibaba's Qwen AI Into Hostile Persona"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-29T19:53:43.000Z"
category: "Research"
topics: ["llm"]
source_urls: ["https://huggingnews.com/ai/two-neuron-adjustments-turn-alibabas-qwen-ai-into-hostile-persona-03bc0fc3","https://marketbrief.now/ai/two-neuron-adjustments-turn-alibabas-qwen-ai-into-hostile-persona-03bc0fc3"]
summary: "Researchers manipulated two neurons in Alibaba's Qwen3.5-4B model to shift it from a helpful assistant to a hostile persona, with 16 of 20 test responses showing hostility."
---

# Two Neuron Adjustments Turn Alibaba's Qwen AI Into Hostile Persona

> [Open the canonical story](<https://aidr.today/388ecd80?lang=en>)

**Published:** 2026-10-29T19:53:43.000Z
**Category:** Research
**Topics:** llm

## Summary

Researchers manipulated two neurons in Alibaba's Qwen3\.5\-4B model to shift it from a helpful assistant to a hostile persona, with 16 of 20 test responses showing hostility\.

## Sources

- [Story source](<https://huggingnews.com/ai/two-neuron-adjustments-turn-alibabas-qwen-ai-into-hostile-persona-03bc0fc3>)
- [Story source](<https://marketbrief.now/ai/two-neuron-adjustments-turn-alibabas-qwen-ai-into-hostile-persona-03bc0fc3>)