DeepSeek AI Fails Multiple Security Tests, Raising Red Flag for Businesses

Summary: Recent testing from AppSOC revealed that the Chinese generative AI model DeepSeek failed 6,400 security tests, showcasing a critical lack of guardrails and a high risk for enterprises. The model demonstrated alarming failure rates in generating malware and viruses, with researchers recommending organizations to refrain from using it for any business applications. Given these results, reliance on DeepSeek poses significant security threats to sensitive data and operational integrity.

Affected: Organizations considering the use of DeepSeek for business applications

Keypoints :

• DeepSeek failed multiple security tests, earning a “high risk” rating of 8.3 out of 10 on the AppSOC scale.
• Failure rates ranged from 19.2% to as high as 98%, particularly dangerous in categories of malware and virus generation.
• AppSOC advises organizations to avoid using DeepSeek in applications handling personal information, sensitive data, or intellectual property.
• Security measures should include auditing and continuous monitoring of AI models used within organizations.