1 report
CrowdStrike research indicates that the DeepSeek-R1 model produces less secure code when prompts contain terms sensitive to Chinese regulations. This unpredictability introduces technical risks, forcing enterprises to increase validation and security scanning for all generated outputs.
