Sign in Sign up
JA · EN · FR
SecPulse
Categories
CVE-2026-64849 CISA KEV critical CVSS 9.3

[KEV] SSRF (Server-Side Request Forgery) in mlflow (CVE-2026-64849)

Summary

SSRF in mlflow (CVE-2026-64849). Confidential information can be exposed externally. Exploitable via POST /api/2.0/mlflow/webhooks/{id}/test. Listed in CISA KEV — actively exploited. Mitigation: upgrade to 3.15.0 or later.

Original description

The description as published by the source.

Summary

The default MLflow Tracking Server (mlflow server, no authentication, default SQLite backend) exposes the model-registry webhooks API unauthenticated, including a synchronous POST /api/2.0/mlflow/webhooks/{id}/test endpoint that returns the upstream response status and body to the caller. The SSRF guard added in PR #20747 (_validate_webhook_url, shipped in 3.10.0) resolves the webhook hostname and rejects non-public IPs, but it is bypassable: delivery follows HTTP redirects (no allow_redirects=False) and never pins the validated IP. An attacker hosts a public HTTPS endpoint that passes the guard and returns 302 Location: http://169.254.169.254/... (or http://127.0.0.1:...); MLflow follows it and never re-validates the redirect target. Because /test reflects the response body, this is an unauthenticated full-read SSRF on a default server.

Details

Three facts combine:

  1. Webhook endpoints are unauthenticated on a default server. The only webhook authorization lives in the optional auth plugin (mlflow/server/auth/__init__.py, WEBHOOK_BEFORE_REQUEST_HANDLERS), which is not loaded by default.

  2. The guard validates but pins nothing — mlflow/utils/validation.py _validate_webhook_url:

schemes = _MLFLOW_WEBHOOK_ALLOWED_SCHEMES.get()        # default ["https"]
if parsed_url.scheme not in schemes: raise ...
if not _MLFLOW_WEBHOOK_ALLOW_PRIVATE_IPS.get():        # default False
    for addr_info in socket.getaddrinfo(hostname, None):…

Source: GitHub Security Advisories (Global) NVD GitHub Advisory CISA KEV

Auto-generated summary

Assembled by SecPulse from the source data (CVSS, CWE, affected versions) using templates — no AI involved. Check the original description and references below for exact details.

A vulnerability tracked as CVE-2026-64849 has been found in mlflow.
Attackers can target a specific entry point like POST /api/2.0/mlflow/webhooks/{id}/test over the network to misuse the product.
Confidential information can be exposed externally. CVSS score: 9.3/10.
Note: CISA has officially confirmed this is actively exploited in the wild. Treat it with elevated urgency.
What to do: upgrade mlflow to 3.15.0 or later.
If unsure, ask your IT team or search "mlflow CVE-2026-64849" on the vendor's site.

CVE-2026-64849 (mlflow) — CWE-918 / CVSS v3 9.3
Attack vector: remote (network-reachable) / unauthenticated / no user interaction
Attack surface: POST /api/2.0/mlflow/webhooks/{id}/test / POST /api/2.0/mlflow/webhooks / POST /api/2.0/mlflow/webhooks/ / _validate_webhook_url
Patched: 3.15.0 — apply immediately
Listed in CISA KEV — actively exploited, treat as top priority.
Plan: 1) Audit SBOM/dependencies, 2) Stage→prod upgrade, 3) Add WAF/proxy monitoring on affected endpoints, 4) Hunt IOCs in logs.
Refs: see the GHSA / vendor advisory / patched release linked on this page.

❓ What is the problem

SSRF (CWE-918) exists in mlflow. Attackers reach the vulnerable code path via POST /api/2.0/mlflow/webhooks/{id}/test without authentication.

📍 Affected scope

Check the vendor advisory for the affected versions of mlflow. Attack surface: POST /api/2.0/mlflow/webhooks/{id}/test / POST /api/2.0/mlflow/webhooks / POST /api/2.0/mlflow/webhooks/ / _validate_webhook_url.

🔥 Severity

Severity: Critical (CVSS 9.3/10). Confidential information can be exposed externally. Listed in CISA KEV — actively exploited in the wild, treat as top priority.

🔧 How to fix

Update to 3.15.0.

🛡️ Workaround

Until the patch is applied: disable the affected feature, apply WAF rules, or restrict access via network ACLs.

🔍 Detection

Search webserver/proxy logs for unusual POST /api/2.0/mlflow/webhooks/{id}/test requests with malformed payloads or SQL meta-characters. Run grep -r 'mlflow' . against your dependency files (package-lock.json, requirements.txt, go.sum) to find affected services.

Response actions

Concrete steps and command examples for SOC/SRE teams to execute in order

  1. 1
    Identify exposure identify 
    grep -r 'mlflow' . | grep -v node_modules

    リポジトリと本番環境の依存ファイル (package-lock.json / requirements.txt / go.sum / Gemfile.lock 等) で mlflow を grep し、稼働しているサービス・バージョンを把握する。

  2. 2
    Hunt for indicators of compromise detect 
    grep 'POST /api/2.0/mlflow/webhooks/{id}/test' /var/log/nginx/access.log | grep -E '(unusual_payload|sqli_pattern)'

    アクセスログで POST /api/2.0/mlflow/webhooks/{id}/test への異常なリクエスト (不正な認証ヘッダ・SQLメタ文字)を過去 30〜90日分捜索。WAF/SIEM があれば該当パスのアラート発火履歴を確認。

  3. 3
    Consider incident declaration escalate 
    Notify SOC / on-call

    CISA KEV登録済 = 実環境で悪用が観測されている。Step 2 で兆候があればインシデント対応宣言、無くてもパッチ適用までWAF強化を最優先で。

  4. 4
    Apply patch patch 
    Upgrade mlflow to 3.15.0

    ステージング環境で 3.15.0 に上げて回帰テスト → 本番反映。回帰テストはアプリの主要ハッピーパスと、Step 2 で見つけた異常検知の続報チェックを含めること。

  5. 5
    Post-deployment verification verify 
    Replay attack against POST /api/2.0/mlflow/webhooks/{id}/test on staging to confirm patch closes the vector

    パッチ適用後、ステージングで PoC または同等の悪用パターンを再現して脆弱性が閉じたことを確認。本番では Step 2 と同じログクエリでアラート再発が無いか継続監視。

Affected packages

pip mlflow
[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"3.15.0"}]}]
PyPI mlflow
[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"3.15.0"}]}]

References

We use cookies for analytics and ad measurement. This setting applies to all Game co., Ltd. sites. Details