Definition
A flaw in a shared AI platform where one customer (tenant) can read, modify, or delete data belonging to a completely separate customer. This happens when the platform incorrectly checks — or fails to check — whether a request is authorised to access a given user's or organisation's data. In AI platforms this can expose chat histories, AI outputs, credentials, and sensitive business data.
Why it matters
Most enterprise AI tools are delivered as shared cloud services; cross-tenant flaws mean a competitor or criminal who also uses the same platform could silently access your AI conversations and the confidential data inside them. Multiple major platforms were found vulnerable in the same reporting window.