Attack  ·  Glossary

IDOR (Insecure Direct Object Reference) in AI platforms

An IDOR flaw occurs when an AI platform lets a user access another user's data simply by changing an identifier (like a number in a URL or API call) — without checking whether the requesting user is actually allowed to see that resource. In AI platforms, this can expose other users' conversation histories, AI agent outputs, uploaded documents, and tool-call logs.
IDORs are among the most common yet preventable flaws in AI platforms — and they can expose highly sensitive data that users share with AI tools in confidence (medical queries, financial data, strategic planning). Multiple AI platforms including Twenty CRM, Open WebUI, and Langflow were found to have IDOR vulnerabilities simultaneously.
References
OWASP API Security Top 10 — API1:2023 Broken Object Level Authorization
Track this in the live feed See how this plays out in real AI security and governance developments.
Open the feed →