No Hackers Required: 10 Shadow AI Leaks Hiding in Plain Sight | UpGuard

This report catalogs ten real incidents where unapproved AI use, misconfigured AI infrastructure, or vendor pipeline failures led to exposure of sensitive corporate IP and personal data (source code, PII, cloud tokens, meeting transcripts), regulatory disclosures, and at least one extortion event; each entry includes a short description, impact, and concrete mitigations (e.g., enterprise AI gateways, CASB, DLP, OAuth consent monitoring, automated PII masking).