CrowdStrike Identifies Security Flaw in China's DeepSeek-R1 AI Model

CrowdStrike Identifies Security Flaw in China's DeepSeek-R1 AI Model

First seen 20 Nov 2025, 21:18 UTC •Securitybrief •26.3

Article Content

Browse articles
ThreatCluster

CrowdStrike's research reveals that DeepSeek-R1, an AI coding tool from China, generates insecure code when prompted with politically sensitive terms, increasing vulnerability rates by nearly 50%. This finding indicates a new supply chain risk for enterprises utilizing AI-powered developer tools, highlighting the impact of bias in large language models.

How this analysis works

Extracted Entities

1 / 2