WeaknessASC-2026-0068Other weakness
Cross-origin and CSRF flaws in AI dashboards
MediumFixed
AI tool dashboards with permissive CORS or no CSRF protection let a malicious page act as a logged-in user, as Langflow CVE-2025-34291 showed.
How to detect it
- Access-Control-Allow-Origin set to any origin with credentials
- State-changing endpoints without CSRF tokens
How to resolve it
- Update to fixed versions
- Restrict CORS to known origins and require CSRF protection
How to investigate
- Check access logs for cross-origin requests
- Review actions taken while users browsed elsewhere