Your RSA-2048 keys break in 2030. Find every one of them before attackers do.
🐍 PyPI

GHSA-gpfh-jvf9-7wg5

HIGH

Use after free / memory leak in `CollectiveReduceV2`

Also known asBIT-tensorflow-2021-41220CVE-2021-41220PYSEC-2021-412PYSEC-2021-629PYSEC-2021-827
Published
Nov 10, 2021
Updated
Mar 13, 2026
Affected
3 pkgs
Patched
3 / 3
Exploits
1 known

EPSS Exploitation Probability

via FIRST.org ↗
0.2%probability of exploitation in next 30 days
Lower Risk10th percentile+0.18%
0.00%0.23%0.47%0.70%0.0%0.2%Dec 25Apr 26Jun 26

EPSS (Exploit Prediction Scoring System) is a daily probability model maintained by FIRST.org. It estimates the likelihood a CVE will be exploited in production environments within the next 30 days, derived from real-world threat intelligence signals.

Blast Radius

3 pkgs affected
🐍tensorflow🐍tensorflow-cpu🐍tensorflow-gpu

Real-time download stats are indexed for npm and PyPI packages. This vulnerability affects PyPI packages — download data is not available via public APIs for these ecosystems.

Description

Impact

The async implementation of CollectiveReduceV2 suffers from a memory leak and a use after free:

import tensorflow as tf
  
tf.raw_ops.CollectiveReduceV2(
  input=[],
  group_size=[-10, -10, -10],
  group_key=[-10, -10],
  instance_key=[-10],
  ordering_token=[],
  merge_op='Mul',
  final_op='Div')

This occurs due to the asynchronous computation and the fact that objects that have been std::move()d from are still accessed:

auto done_with_cleanup = [col_params, done = std::move(done)]() {
  done();
  col_params->Unref();
};
OP_REQUIRES_OK_ASYNC(c,
                     FillCollectiveParams(col_params, REDUCTION_COLLECTIVE,
                                          /*group_size*/ c->input(1),
                                          /*group_key*/ c->input(2),
                                          /*instance_key*/ c->input(3)),
                     done);

Here, done is already moved from by the time OP_REQUIRES_OK_ASYNC macro needs to invoke it in case of errors. In this case, we get an undefined behavior, which can manifest via crashes, std::bad_alloc throws or just memory leaks.

Patches

We have patched the issue in GitHub commit ca38dab9d3ee66c5de06f11af9a4b1200da5ef75.

The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, as this version is the only one that is also affected.

For more information

Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.

Attribution

This vulnerability has been reported by members of the Aivul Team from Qihoo 360.

Affected Packages

3 total 3 fixed
EcosystemPackageVulnerable rangeFix
🐍PyPItensorflow2.6.0&&< 2.6.12.6.1
🐍PyPItensorflow-cpu2.6.0&&< 2.6.12.6.1
🐍PyPItensorflow-gpu2.6.0&&< 2.6.12.6.1
Exploits & PoCs
1

Research use only. For defensive security, authorized penetration testing, and academic research only. Never execute exploit code against systems without explicit written authorization.

Detection & mitigation playbook

Open-source dependency
  1. Detect

    Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for tensorflow. O3's reachability analysis confirms whether the vulnerable code path is actually invoked in your application, so you act on real exposure instead of every transitive match.

  2. Fix

    Update tensorflow to 2.6.1 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms GHSA-gpfh-jvf9-7wg5 is resolved across your whole dependency graph.

  3. Workarounds

    If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.

  4. How O3 protects you

    O3 pinpoints whether GHSA-gpfh-jvf9-7wg5 is reachable in your code and exactly where to fix it, then blocks exploitation in production at runtime until the patched version is deployed.

Tailored to GHSA-gpfh-jvf9-7wg5. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.

Frequently Asked Questions

### Impact The [async implementation](https://github.com/tensorflow/tensorflow/blob/8d72537c6abf5a44103b57b9c2e22c14f5f49698/tensorflow/core/kernels/collective_ops.cc#L604-L615) of `CollectiveReduceV2` suffers from a memory leak and a use after free: ```python import tensorflow as tf tf.raw_ops.CollectiveReduceV2( input=[], group_size=[-10, -10, -10], group_key=[-10, -10], instance_key=[-10], ordering_token=[], merge_op='Mul', final_op='Div') ``` This occurs due to the asynchronous computation and the fact that objects that have been `std::move()`d from are still accessed:
O3 Security · Impact-Aware SCA

Is GHSA-gpfh-jvf9-7wg5 in your dependencies?

O3 detects GHSA-gpfh-jvf9-7wg5 across PyPI dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.