Researchers: DeepSeek's R1 failed to detect or block any of the 50 malicious prompts that were tested; Adversa: R1 is vulnerable to many jailbreaking tactics (Wired)

5 days ago 4

Wired:
Researchers: DeepSeek's R1 failed to detect or block any of the 50 malicious prompts that were tested; Adversa: R1 is vulnerable to many jailbreaking tactics  —  Security researchers tested 50 well-known jailbreaks against DeepSeek's popular new AI chatbot.  It didn't stop a single one.

1. Homepage
2. Technology
3. Researchers: DeepSeek's R1 failed to detect or block any of the 50 malicious prompts that were tested; Adversa: R1 is vulnerable to many jailbreaking tactics (Wired)