DeepSeek AI Models Are Easier to Hack Than US Rivals, Warn Researchers
ID: 4517d53e-9273-529d-978d-bd7c229a6fa1
STIX ID: report--4517d53e-9273-529d-978d-bd7c229a6fa1
Feed Name: TechRepublic Security
Threat Score
NIST/CAISI evaluated DeepSeek models (V3.1, R1, R1-0528) and found them substantially weaker than leading U.S. models on safety, accuracy, and security: jailbreak prompts yielded harmful outputs in 95–100% of tests, agent-hijack tests exfiltrated two-factor codes in 37% of cases, and the models underperformed on engineering/cybersecurity benchmarks; despite rapid adoption, these built-in vulnerabilities raise a significant abuse risk.
Your team is not currently subscribed to this feed. You must subscribe to it in order to see this post.