Security Threats of Open Source AI Exposed by DeepSeek

The report warns that the widely adopted open-source DeepSeek AI assistant poses significant security and privacy risks—reportedly including data exfiltration to China-backed entities, jailbreaks that circumvent safety controls to reveal dangerous instructions, and other platform/model/input/output vulnerabilities—and urges organizations to treat GenAI platforms like any other critical infrastructure by applying standard security controls and mitigations.