---
format: "aidr-story-markdown/v1"
id: "ecb59480543c9123de85e471f35de7d7386346caf6679d43c08b2e0be1ca3a72"
canonical_url: "https://aidr.today/ecb59480?lang=en"
title: "Free Chinese AI Models Are Getting Scary Good at Hacking"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-06T18:43:43.000Z"
category: "Research"
topics: ["deepseek","safety"]
source_urls: ["https://www.wired.com/story/free-chinese-ai-models-are-getting-scary-good-at-hacking/"]
summary: "An experiment showed that the free Chinese AI models DeepSeek and GLM demonstrated hacking skills close to OpenAI's most advanced non-public cyber model when attempting to break into simulated networks."
---

# Free Chinese AI Models Are Getting Scary Good at Hacking

> [Open the canonical story](<https://aidr.today/ecb59480?lang=en>)

**Published:** 2026-11-06T18:43:43.000Z
**Category:** Research
**Topics:** deepseek, safety

## Summary

An experiment showed that the free Chinese AI models DeepSeek and GLM demonstrated hacking skills close to OpenAI's most advanced non\-public cyber model when attempting to break into simulated networks\.

## Sources

- [Story source](<https://www.wired.com/story/free-chinese-ai-models-are-getting-scary-good-at-hacking/>)