CVE intelligence and bounded remediation
CVE-2026-2651 — Lfprojects Mlflow security vulnerability
A vulnerability in MLflow versions <=3.10.1.dev0 allows unauthorized access to multipart upload (MPU) endpoints when the `--serve-artifacts` mode is enabled. The authorization logic does not enforce resource-level permission checks for `/mlflow-artifacts/mpu/*` endpoints, enabling attackers to overwrite artifacts belonging to other users. This can lead to unauthorized cross-user writes, model supply chain poisoning, and arbitrary code execution when compromised models are loaded. The issue is resolved in version 3.10.0.
- Severity
- Critical
- CVSS
- 9 (3.1)
- Published
- 2026-05-25
- CISA KEV
- Not currently listed
- Ecosystem
- software/application
- Weaknesses
- CWE-862, CWE-1220
Affected products
- lfprojects / mlflow
Matched remediation archetype
Authorization bypass, IDOR, and cross-tenant access
This catalog composition supplies bounded fallback guidance. Explicitly reviewed curated workflows load with the complete record below.
Check exposure
- Map object and action authorization checks across API, UI, batch, import/export, and background-job paths.
- Identify tenant, ownership, role, and policy boundaries for affected resources and administrative operations.
- Use synthetic fixtures to compare intended access matrices without accessing another user's real data.
Remediate safely
- Enforce server-side authorization at each resource access and state transition using the authenticated principal and trusted tenant context.
- Scope data queries by tenant and ownership; treat client-supplied identifiers, roles, and policy claims as untrusted.
- Add deny-by-default policy tests for horizontal and vertical access across every affected transport.
Authoritative sources
Complete CVE record and remediation plan
The detailed catalog view below loads this exact record, its source evidence, and the full seven-phase agentic change plan.