Source: The Next Web
Z.ai's decision to encrypt user workspaces in a way that only the company can decrypt—not even users themselves—inverts the typical SaaS trust model and creates a single point of failure for data verification. This is a control mechanism that prevents independent audit of what the company does with uploaded content, making deletion claims unverifiable and shifting all data sovereignty to the vendor. For consumers evaluating AI tools handling sensitive work, this represents a structural red flag that no privacy policy can overcome.