LiveActive security incident?Get immediate response
CVE Record

CVE-2026-2614: Arbitrary File Read via Prompt Tag Source Validation Bypass in mlflow/mlflow

A vulnerability in the `_create_model_version()` handler of `mlflow/server/handlers.py` in mlflow/mlflow versions 3.9.0 and earlier allows an unauthenticated remote attacker to read arbitrary files from the server's filesystem. The issue arises when a `CreateModelVersion` request includes the tag `mlflow.prompt.is_prompt`, which bypasses source path validation. This enables an attacker to store an arbitrary local filesystem path as the model version source. The `get_model_version_artifact_handler()` function later uses this source to serve files without verifying the model version's prompt status, leading to a complete confidentiality compromise. This issue is fixed in version 3.10.0.

HighCVSS 7.5Not KEV-listedUpdated
Discuss exposureBack to CVEs
Glexia's TakeAutomated analysishigh

Security readout for executives and security teams

Plain-English summary

MLflow 3.9.0 and earlier can expose sensitive files from the server hosting it. A remote attacker needs no account or user interaction. Successful abuse affects confidentiality rather than modifying data or disrupting service, but exposed files could include configuration, credentials, or proprietary information.

Executive priority

Treat externally reachable or broadly accessible MLflow servers as urgent because exploitation needs no credentials and may disclose sensitive server files. Upgrade to 3.10.0 or later promptly. Accelerate remediation for internal instances as well, particularly where MLflow can read credentials, configuration, intellectual property, or regulated data.

Technical view

A crafted CreateModelVersion request containing the mlflow.prompt.is_prompt tag bypasses source-path validation in _create_model_version(). MLflow can consequently retain a local filesystem path as the model source, which get_model_version_artifact_handler() later serves without confirming prompt status. CVSS 3.0 is 7.5, reflecting network access, low complexity, no privileges, and high confidentiality impact.

Likely exposure

Deployments running MLflow 3.9.0 or earlier are affected, with greatest risk where the MLflow service is reachable from untrusted networks. Exposure depends on network accessibility and which server files the MLflow process can read. The supplied evidence does not identify affected deployment counts or specific configurations that eliminate the flaw.

Exploitation context

The source bundle marks this CVE as absent from KEV, and no supplied source confirms active exploitation. Nevertheless, exploitation is described as remote, unauthenticated, low complexity, and requiring no user interaction. Public vulnerability details and the corrective commit are available, increasing the importance of timely remediation.

Researcher notes

The reported weakness is CWE-22 and involves inconsistent validation across two handlers: model-version creation accepts an attacker-controlled local source under the prompt tag, while artifact retrieval trusts that stored source without rechecking prompt status. The supplied sources identify 3.10.0 as fixed but provide no evidence of exploitation in the wild.

Mitigation direction

  • Upgrade every affected MLflow deployment to version 3.10.0 or later.
  • Until upgraded, restrict MLflow network access to trusted users and follow current vendor guidance.
  • Review Red Hat advisories for packaged MLflow components and apply applicable updates.
  • If suspicious access is confirmed, assess and rotate secrets stored in readable server files.

Validation and detection

  • Inventory MLflow versions and identify every deployment running version 3.9.0 or earlier.
  • Determine whether affected services are reachable from untrusted networks without authentication.
  • Review model-version records and request logs for unexpected prompt tags or local filesystem source paths.
  • After upgrading, verify version 3.10.0 or later and test normal registry and artifact operations.
  • Investigate potential access to sensitive files if suspicious records or requests are discovered.
Prepared
Confidence
high
Sources
7

Public sources used

Generated from the cited source records. This long-tail analysis has not been individually reviewed by a named human.

Potential ATT&CK relevance

Conservative CVE-to-ATT&CK context

These mappings and lookup hints may be relevant to the vulnerability behavior, CWE, affected product, or exposure path. Glexia-inferred context is not an official MITRE, ATT&CK, CWE, or CVE Program mapping.

ATT&CK lookup starting points

Use these exact CWE pages and searches to review the Glexia ATT&CK library from this CVE's weakness and description context.

cwe · medium confidence lookup

CWE-22: File access and web shell behavior lookup

File traversal and upload weaknesses can lead teams to review file, web shell, execution, and collection telemetry. Open the exact CWE lookup page first, then review the ATT&CK searches from that MITRE weakness context. This is a Glexia lookup hint, not an official ATT&CK mapping.

Open ATT&CK lookup
description · low confidence lookup

File access behavior lookup

The CVE wording references file access or upload behavior, so file telemetry and web shell review may help. This is a Glexia inferred lookup path, not an official MITRE, ATT&CK, or CVE Program mapping.

Open ATT&CK lookup
cve · low confidence lookup

CVE-2026-2614 mapping review

Open the CVE-to-ATT&CK bridge for reviewed, inferred, or future official mappings tied to this CVE.

Open ATT&CK lookup
Vulnerability profileCVE Program record
Severity
High
CVSS
7.5 (3.0)
Known Exploited
No
Published

Vector: CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N

Official CVE source material

CNA and ADP enrichment extracted from CVE v5

These fields come from the CVE record and ADP containers, not from Glexia's Take. They preserve time-varying source decisions such as CISA SSVC, KEV status, CVSS metrics, and provider references.

2CVSS vectors
5Timeline events
2ADP providers
9Source links

SSVC decision data

CISA-ADPCISA Coordinator
Timestamp
Version
2.0.3
Exploitation: pocAutomatable: yesTechnical Impact: partial

CVSS vector scores

2 official scores

We collect every scored CVSS vector available in the official CNA and ADP containers. When more than one version is present, the table keeps the source vectors side by side instead of collapsing them into the highest score.

ScoreVersionSeverityVectorExploitImpactSource
7.5CVSS 3.0HighCVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N3.93.6@huntr_ai
7.5CVSS 3.1HighCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N3.93.6redhat-SADP

Vulnerability scoring details

Base CVSS 3.1 score

7.5High
CVSS 3.1 vector shape for CVE-2026-2614Attack VectorAttack ComplexityPrivileges RequiredUser InteractionScopeConfidentiality ImpactIntegrity ImpactAvailability Impact

Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N

Attack Vector
NetworkAdjacentLocalPhysical
Attack Complexity
LowHigh
Privileges Required
NoneLowHigh
User Interaction
NoneRequired
Scope
ChangedUnchanged
Confidentiality Impact
HighLowNone
Integrity Impact
HighLowNone
Availability Impact
HighLowNone

Vulnerability timeline

Timeline events are normalized from CVE metadata, CNA source timelines, ADP timelines, and KEV metadata when present.

  1. CVE reservedCVE Program

    The CVE ID was reserved by the assigning CNA.

  2. ADP timelineredhat-SADP

    Made public.

  3. CVE publishedCVE Program

    The CVE record was published.

  4. ADP timelineredhat-SADP

    Reported to Red Hat.

  5. CVE updatedCVE Program

    The CVE record metadata indicates this as the latest update time.

ADP provider summaries

CISA-ADPCISA ADP Vulnrichment
other:ssvc
redhat-SADPmlflow: mlflow: Arbitrary file read via bypassed source path validation
other:Red Hat severity ratingcvssV3_1
  • 2026-05-11T20:05:23.444Z: Reported to Red Hat.
  • 2026-05-11T19:02:46.025Z: Made public.

Source materials

Affected products

Products and packages named in the record

VendorProductVersion / packageStatus
mlflowmlflow/mlflowunspecifiedListed
Weakness

CWE details

CWE-22

CWE links open Glexia weakness intelligence pages with official CWE context, developer remediation guidance, and related CVE mappings.

CWE-22 · source CWE mapping

Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') represents a recurring weakness pattern that can create exploitable paths when design, validation, or implementation controls are missing.

Open CWE intelligenceSame-CWE CVEs