ChalkyC
@chalkyc.bsky.social
Reposted by ChalkyC
NEW: Security researchers tested 50 well-known jailbreaks against DeepSeek’s popular new AI chatbot. It didn’t stop a single one.
DeepSeek’s Safety Guardrails Failed Every Test Researchers Threw at Its AI Chatbot
Security researchers tested 50 well-known jailbreaks against DeepSeek’s popular new AI chatbot. It didn’t stop a single one.
www.wired.com
January 31, 2025 at 6:39 PM
NEW: Security researchers tested 50 well-known jailbreaks against DeepSeek’s popular new AI chatbot. It didn’t stop a single one.