---
format: "aidr-story-markdown/v1"
id: "2f9740f31c6734ed071e20f65f25bd716f05410b843891db9d1f8e96d7086fff"
canonical_url: "https://aidr.today/2f9740f3?lang=en"
title: "Chinese Hackers Adopt DeepSeek to Scale Global Attacks via Weak AI Guardrails"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-08-24T20:51:36.000Z"
category: null
topics: ["deepseek","openai","anthropic","claude-code","safety"]
source_urls: ["https://huggingnews.com/cybersecurity/chinese-hackers-adopt-deepseek-to-scale-global-attacks-via-weak-ai-guard-4c5d6f09","https://x.com/business/status/2091973862518816905","https://x.com/chigrl/status/2091966616611287140","https://x.com/Polymarket/status/2091965514839200075","https://x.com/financialjuice/status/2091937099947626698","https://x.com/FirstSquawk/status/2091980387996389565","https://x.com/negligible_cap/status/2091938654289567872"]
summary: "Team T5 researchers found that China-linked cyber groups are leveraging low-cost artificial intelligence to increase the frequency and success of their intrusions. The DeepSeek model has become the primary tool for these actors due to its minimal safety restrictions, which are significantly easier to navigate than the strict guardrails found in Western AI models. Beyond open-source tools, China-backed spies have employed chatbots from OpenAI and Anthropic to facilitate system intrusions. CyCraft reported that ChatGPT was used in an attack on a Western think tank, while the group Slime22 bypassed safety controls in Anthropic's Claude Code by posing as an engineer to conduct lateral movements inside a Taiwanese technology company."
---

# Chinese Hackers Adopt DeepSeek to Scale Global Attacks via Weak AI Guardrails

> [Open the canonical story](<https://aidr.today/2f9740f3?lang=en>)

**Published:** 2026-08-24T20:51:36.000Z
**Topics:** deepseek, openai, anthropic, claude\-code, safety

## Summary

Team T5 researchers found that China\-linked cyber groups are leveraging low\-cost artificial intelligence to increase the frequency and success of their intrusions\. The DeepSeek model has become the primary tool for these actors due to its minimal safety restrictions, which are significantly easier to navigate than the strict guardrails found in Western AI models\. Beyond open\-source tools, China\-backed spies have employed chatbots from OpenAI and Anthropic to facilitate system intrusions\. CyCraft reported that ChatGPT was used in an attack on a Western think tank, while the group Slime22 bypassed safety controls in Anthropic's Claude Code by posing as an engineer to conduct lateral movements inside a Taiwanese technology company\.

## Sources

- [Story source](<https://huggingnews.com/cybersecurity/chinese-hackers-adopt-deepseek-to-scale-global-attacks-via-weak-ai-guard-4c5d6f09>)
- [Story source](<https://x.com/business/status/2091973862518816905>)
- [Supporting source](<https://x.com/chigrl/status/2091966616611287140>)
- [Supporting source](<https://x.com/Polymarket/status/2091965514839200075>)
- [Supporting source](<https://x.com/financialjuice/status/2091937099947626698>)
- [Supporting source](<https://x.com/FirstSquawk/status/2091980387996389565>)
- [Supporting source](<https://x.com/negligible_cap/status/2091938654289567872>)