Sign in
Deutsch English Español Polski
High

Apache Airflow Vulnerable to Deserialization of Untrusted Data

  • Disclosed: 01.06.2026
  • Updated: 11.10.2026 15:17

Apache Airflow's scheduler-side deadline-reference decoder (SerializedCustomReference.deserialize_reference) imported and dispatched arbitrary class paths drawn from DAG-author-controlled serialized state without an allowlist or plugin-registry gate. A DAG author whose code reaches the scheduler — the default on single-host deployments where the DAG bundle is importable from the scheduler process — could embed a custom DeadlineReference whose serialized form named an attacker-controlled module path, causing the scheduler to import_string(...) and instantiate that class with a live SQLAlchemy session attached. Affects deployments where DAG-author code is less trusted than the scheduler process. Users are advised to upgrade to apache-airflow 3.2.2 or later.

Affected software

Click a technology to see all vulnerabilities affecting it.

PlatformPackageVulnerable versionsFixed in
Python (PyPI) apache-airflow >= 3.2.0, < 3.2.2 3.2.2

References

CVSS score 7.3/10
CVSS vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L
Weakness (CWE)
CWE-502
Platform
Python (PyPI)
Package
apache-airflow
Disclosed
01.06.2026
Updated
11.10.2026 15:17

Sources and licenses

OSV / GitHub Advisory Database - PYSEC-2026-186 - PyPA Advisory Database; CC BY 4.0

OSV / GitHub Advisory Database - GHSA-2r5m-76wx-56gx - CC BY 4.0

Vulnerability history: apache-airflow

All published vulnerabilities of this package (Python (PyPI)), newest first. Shows whether this is a one-off slip or a package that needs watching.

All (158) in the archive
Total entries
158
First → latest
08.2018 09.2026
Without a fix
4
Exploited in attacks
2

Severity breakdown

  • Critical 16
  • High 49
  • Medium 86
  • Low 7

Disclosures per year

Severity:
  1. Apache Airflow: the Core API logout endpoint revokes only a session token presented as the _token cookie. When a client logs out presenting its credential as an Authorization bearer header instead, th
    Critical · 9.1 CVE-2026-86473 >= 3.0.0, < 3.3.2 3.3.2
  2. Apache Airflow's Backfill API authorized a request against a Dag id supplied by the caller whenever the `backfill_id` path segment failed to parse. The authorization dependency parsed it with `int()`
    High · 7.5 CVE-2026-68968 < 3.3.1 3.3.1
  3. Apache Airflow's Task SDK rebuilt a `Callback` object from serialized data by re-running its constructor, which imports the module named by the stored callback path. Because `SyncCallback` is itself a
    High · 8.8 CVE-2026-67587 >= 3.3.0, < 3.3.1 3.3.1
  4. Apache Airflow 3.3.0 moved human-in-the-loop tasks from the triggerer to a new `awaiting_input` task state swept by the scheduler. That sweep deserializes the task instance's `next_kwargs` without an
    High · 7.3 CVE-2026-67260 >= 3.3.0, < 3.3.1 3.3.1
  5. Apache Airflow's serialization layer reconstructed exception nodes by calling `import_string()` on a class name taken from the serialized blob and instantiating it with arguments from the same blob, w
    High · 8.8 CVE-2026-58076 >= 3.3.0, < 3.3.1 3.3.1
  6. A bug in `BaseSerialization.deserialize()` allowed unrestricted `import_string()` of attacker-controlled class paths when the Scheduler / API Server loaded a serialized DAG: a DAG author could embed a
    Critical · 9.8 CVE-2026-33264 < 3.3.0 3.3.0
  7. A bug in Apache Airflow's XCom PATCH endpoint `PATCH /api/v2/xcomEntries/{key}` allowed an authenticated UI/API user with XCom write permission on a Dag to set XCom entries under reserved key names (e
    High · 8.8 CVE-2026-42359 >= 3.2.0, < 3.2.2 3.2.2
  8. A bug in Apache Airflow's bulk Task Instances API (`PATCH/DELETE /api/v2/dags/{dag_id}/dagRuns/{dag_run_id}/taskInstances`) evaluated authorization against the `dag_id` resolved from the URL path whil
    High · 7.5 CVE-2026-41084 >= 3.2.0, < 3.2.2 3.2.2
  9. Apache Airflow's official documentation at `core-concepts/dag-run.html` ("Passing Parameters when triggering Dags") showed a verbatim `BashOperator(bash_command="echo value: {{ dag_run.conf['conf1'] }
    Critical · 9.1 CVE-2026-42252 >= 3.0.0, < 3.2.2 3.2.2
  10. Apache Airflow: Authenticated users can bypass the `is_safe_url` check
    High · 7.2 CVE-2026-40961 < 3.2.2 3.2.2
  11. Apache Airflow Vulnerable to Deserialization of Untrusted Data this entry
    High · 7.3 CVE-2026-45360 >= 3.2.0, < 3.2.2 3.2.2
  12. Apache Airflow allows code execution through crafted XCom payloads
    High · 7.2 CVE-2026-25917 < 3.2.0 3.2.0
  13. Apache Airflow: JWT token appearing in logs
    High · 7.5 CVE-2026-31987 >= 3.0.0, < 3.2.0 3.2.0
  14. Apache Airflow: RCE by race condition in example_xcom dag
    High · 8.1 CVE-2025-54550 < 3.2.0 3.2.0
  15. Dag Authors, who normally should not be able to execute code in the webserver context could craft XCom payload causing the webserver to execute arbitrary code. Since Dag Authors are already highly tru
    High · 8.8 CVE-2026-33858 >= 3.1.8, < 3.2.0 3.2.0
  16. Apache Airflow: JWT token still valid after logout
    Critical · 9.1 CVE-2025-57735 >= 3.0.0, < 3.2.0 3.2.0
  17. Apache Airflow versions 3.1.0 through 3.1.7 missing authorization vulnerability in the Execution API's Human-in-the-Loop (HITL) endpoints that allows any authenticated task instance to read, approve,
    High · 8.1 CVE-2026-30911 >= 3.1.0, < 3.1.8 3.1.8
  18. Apache Airflow versions 3.1.0 through 3.1.7 session token (_token) in cookies is set to path=/ regardless of the configured [webserver] base_url or [api] base_url.
    High · 7.5 CVE-2026-28779 >= 3.0.0, < 3.1.8 3.1.8
  19. Apache Airflow vulnerable to Code Injection in the web-server context via LogTemplate table
    High · 8.4 CVE-2024-56373 < 2.11.1 2.11.1
  20. Apache Airflow proxy credentials for various providers might leak in task logs
    High · 7.5 CVE-2025-68675 >= 3.0.0b1, < 3.1.6, < 2.11.1 3.1.6, 2.11.1
  21. Apache Airflow secrets in rendered templates could contain parts of sensitive values when truncated
    High · 7.5 CVE-2025-68438 >= 3.1.0, < 3.1.6 3.1.6
  22. Apache Airflow Providers Edge3 exposes internal API allowing RCE in web server context
    Critical · 9.8 CVE-2025-67895 < 2.0.0 2.0.0
  23. Apache Airflow: Sensitive configuration values are not masked in the logs by default
    High · 7.5 CVE-2024-45784 < 2.10.3 2.10.3
  24. Apache Airflow vulnerable to Improper Encoding or Escaping of Output
    High · 8.8 CVE-2024-45498 >= 2.10.0, < 2.10.1 2.10.1
  25. Apache Airflow vulnerable to Execution with Unnecessary Privileges
    High · 8.8 CVE-2024-45034 < 2.10.1 2.10.1
  26. Apache Airflow has DAG Author Code Execution possibility in airflow-scheduler
    High · 8.8 CVE-2024-39877 >= 2.4.0, < 2.9.3 2.9.3
  27. Apache Airflow, versions before 2.8.1, have a vulnerability that allows a potential attacker to poison the XCom data by bypassing the protection of "enable_xcom_pickling=False" configuration setting r
    High · 7.5 CVE-2023-50943 < 2.8.1 2.8.1
  28. Apache Airflow, versions before 2.8.1, have a vulnerability that allows an authenticated user to access the source code of a DAG to which they don't have access. This vulnerability is considered low s
    High · 6.5 CVE-2023-50944 < 2.8.1 2.8.1
  29. Apache Airflow, versions before 2.7.3, has a vulnerability that allows an authorized user who has access to read specific DAGs only, to read information about task instances in other DAGs.  This is a
    High · 6.5 CVE-2023-42781 < 2.7.3 2.7.3
  30. Apache Airflow Celery provider Insertion of Sensitive Information into Log File vulnerability
    High · 7.5 CVE-2023-46215 >= 1.10.0, < 2.7.0 2.7.0
  31. Apache Airflow, versions before 2.7.1, is affected by a vulnerability that allows authenticated users who have access to see the task/dag in the UI, to craft a URL, which could lead to unmasking the s
    High · 6.5 CVE-2023-40712 < 2.7.1 2.7.1
  32. Apache Airflow, in versions prior to 2.7.0, contains a security vulnerability that can be exploited by an authenticated user possessing Connection edit privileges. This vulnerability allows the user t
    High · 8.1 CVE-2023-37379 < 2.7.0 2.7.0
  33. The session fixation vulnerability allowed the authenticated user to continue accessing Airflow webserver even after the password of the user has been reset by the admin - up until the expiry of the s
    High · 8.0 CVE-2023-40273 < 2.7.1rc1 2.7.1rc1
  34. apache-airflow-providers-apache-drill Improper Input Validation vulnerability
    High · 7.5 CVE-2023-39553 < 2.4.3 2.4.3
  35. Execution with Unnecessary Privileges, : Exposure of Sensitive Information to an Unauthorized Actor vulnerability in Apache Software Foundation Apache Airflow.The "Run Task" feature enables authentica
    High · 8.8 CVE-2023-39508 < 2.6.0 2.6.0
  36. Apache Airflow, versions before 2.6.3, is affected by a vulnerability that allows an unauthorized actor to gain access to sensitive information in Connection edit view. This vulnerability is considere
    High · 6.5 CVE-2022-46651 < 2.6.3 2.6.3
  37. Apache Airflow, versions before 2.6.3, is affected by a vulnerability that allows unauthorized read access to a DAG through the URL. It is recommended to upgrade to a version that is not affected
    High · 6.5 CVE-2023-35908 < 2.6.3 2.6.3
  38. Apache Airflow, versions before 2.6.3, is affected by a vulnerability that allows an attacker to cause a service disruption by manipulating the run_id parameter. This vulnerability is considered low s
    High · 6.5 CVE-2023-22888 < 2.6.3 2.6.3
  39. Apache Airflow, versions before 2.6.3, has a vulnerability where an authenticated user can use crafted input to make the current request hang. It is recommended to upgrade to a version that is not aff
    High · 6.5 CVE-2023-36543 < 2.6.3 2.6.3
  40. Apache Airflow, versions before 2.6.3, is affected by a vulnerability that allows an attacker to perform unauthorized file access outside the intended directory structure by manipulating the run_id pa
    High · 6.5 CVE-2023-22887 < 2.6.3 2.6.3
Show more (20) Showing 40 of 65

Know which vulnerabilities affect your application

With the DockRay integration and software inventory enabled, DockRay checks reported component versions against known vulnerabilities and notifies you about confirmed matches. Coverage depends on the supported technology and available advisory data.

Connect your application
DockRay newsletter

Application security in your inbox

Important vulnerabilities, actionable guidance and practical advice for your applications. Choose your technologies and follow what matters to your project.

  • A daily digest of new vulnerabilities, critical ones included.
  • Notifications about important updates.
  • Practical advice on application security and monitoring.

Leave empty to receive updates for every technology.

Privacy policy

Free to subscribe. Unsubscribe with one click in any email.