Advisory record · GHSA-vxv8-r8q2-63xw

TensorFlow vulnerable to `CHECK` fail in `FractionalMaxPoolGrad`

CVSS 5.9 MediumEPSS 0.0049 · 42nd percentile

Published by the GitHub Advisory Database at 16 Sept 2022, 22:26 UTC. agentprime has not analysed this yet, so whether it affects any app is not known.

PackageEcosystemAffectedFixed in
tensorflowPyPI<2.7.2 || >=2.8.0, <2.8.1 || >=2.9.0, <2.9.12.7.2, 2.8.1, 2.9.1
tensorflow-cpuPyPI<2.7.2 || >=2.8.0, <2.8.1 || >=2.9.0, <2.9.12.7.2, 2.8.1, 2.9.1
tensorflow-gpuPyPI<2.7.2 || >=2.8.0, <2.8.1 || >=2.9.0, <2.9.12.7.2, 2.8.1, 2.9.1
Identifiers
GHSA-vxv8-r8q2-63xw · BIT-tensorflow-2022-35981 · CVE-2022-35981 · PYSEC-2026-1041 · PYSEC-2026-3250 · PYSEC-2026-3375
Weakness (CWE)
CWE-617
Published
16 Sept 2022
Updated
13 Jul 2026

Machine-written summary

Summary written by Claude Haiku from the source advisory. Not reviewed by a person.

TensorFlow (packages tensorflow, tensorflow-cpu, and tensorflow-gpu) versions before 2.7.2, from 2.8.0 before 2.8.1, and from 2.9.0 before 2.9.1 are affected. The FractionalMaxPoolGrad operation uses CHECK failures instead of returned errors when given incorrectly sized inputs, so an attacker who can supply such inputs can crash the process as a denial of service. Upgrade to 2.7.2, 2.8.1, 2.9.1, or later.

Written 6 Nov 2026.

Not analysed yet

We haven’t published our analysis of this Advisory yet. Once a person has reviewed it, this page will say what’s vulnerable, what has to be true for your app to be affected, and how to check.

Severity and exploitation

ScoreRatingCVSSVectorSource
5.9 Medium 3.1 CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H CNA (GitHub_M)
  • CISA KEV Not listed.
  • EPSS 0.0049, the 42nd percentile, on 19 Nov 2026: FIRST’s estimate of the chance of exploitation in the next 30 days.

Timeline

  1. 31 Oct 2026, 15:03 UTC First seen by agentprime
  2. 31 Oct 2026, 15:03 UTC Confirmed by OSV

Sources

Facts on this page come from the GitHub Advisory Database (CC-BY 4.0), OSV and the other sources listed, as published. We don’t reproduce full advisory descriptions, which can contain exploit code.