Skip to content
LLMMedium

CVE-2021-37672: TensorFlow is an end-to-end open source platform for machine learning. In affected versions an attac…

Occurred
Aug 12, 2021
Published
Aug 12, 2021
Source
NVD CVE

Summary

TensorFlow is an end-to-end open source platform for machine learning. In affected versions an attacker can read from outside of bounds of heap allocated data by sending specially crafted illegal arguments to `tf.raw_ops.SdcaOptimizerV2`. The [implementation](https://github.com/tensorflow/tensorflow/blob/460e000de3a83278fb00b61a16d161b1964f15f4/tensorflow/core/kernels/sdca_internal.cc#L320-L353) does not check that the length of `example_labels` is the same as the number of examples. We have patched the issue in GitHub commit a4e138660270e7599793fa438cd7b2fc2ce215a6. The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range.

Tags

  • CVE
  • CVE-2021-37672
  • cvss:MEDIUM

Classification

Category confidence: 85%
Severity confidence: 90%

Rationale: CVE describes a memory safety vulnerability in TensorFlow ML framework, not a direct AI safety incident.

Classifier: Qwen/Qwen3.5-35B-A3B-FP8

Visit source