A report from New York-based cybersecurity firm Wiz revealed that scans of DeepSeek's infrastructure uncovered unsecured data, including digital software keys and chat logs containing user prompts sent to its AI assistant.