CVE-2021-37678 Arbitrary code execution due to YAML deserialization

Critical 9.3 Published Updated

Attackers can execute arbitrary code by tricking TensorFlow into deserializing malicious Keras models in YAML format.

Exploited in the wild
Not known
Login needed
Not needed
Attackable
Local or same network only
Victim action needed
Not needed
Fix
Available

What is CVE-2021-37678?

CVE-2021-37678 is a critical unsafe deserialization vulnerability in Google Tensorflow, published on 12 Aug 2021. It can only be exploited from the same machine or local network, with no action from the victim.

Official description

TensorFlow is an end-to-end open source platform for machine learning. In affected versions TensorFlow and Keras can be tricked to perform arbitrary code execution when deserializing a Keras model from YAML format. The [implementation](https://github.com/tensorflow/tensorflow/blob/460e000de3a83278fb00b61a16d161b1964f15f4/tensorflow/python/keras/saving/model_config.py#L66-L104) uses `yaml.unsafe_load` which can perform arbitrary code execution on the input. Given that YAML format support requires a significant amount of work, we have removed it for now. We have patched the issue in GitHub commit 23d6383eb6c14084a8fc3bdf164043b974818012. 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.

Weakness: CWE-502

Which versions are affected by CVE-2021-37678?

22 open source packages ship an affected version, including tensorflow, tensorflow-cpu and tensorflow-gpu. It affects 1 product.

PackageAffectedFixed in
tensorflow BitnamiFrom 2.3.02.3.4, 2.4.3, 2.5.1
tensorflow PyPIAll versions2.3.4
tensorflow PyPIAll versions2.3.4
tensorflow PyPIFrom 2.3.02.3.4, 2.4.3
tensorflow PyPIFrom 2.4.02.4.3
tensorflow PyPIFrom 2.4.02.4.3
16 more packages
tensorflow PyPIFrom 2.5.02.5.1
tensorflow PyPIFrom 2.5.02.5.1
tensorflow-cpu PyPIAll versions2.3.4
tensorflow-cpu PyPIAll versions2.3.4
tensorflow-cpu PyPIFrom 2.3.02.3.4, 2.4.3
tensorflow-cpu PyPIFrom 2.4.02.4.3
tensorflow-cpu PyPIFrom 2.4.02.4.3
tensorflow-cpu PyPIFrom 2.5.02.5.1
tensorflow-cpu PyPIFrom 2.5.02.5.1
tensorflow-gpu PyPIAll versions2.3.4
tensorflow-gpu PyPIAll versions2.3.4
tensorflow-gpu PyPIFrom 2.3.02.3.4, 2.4.3
tensorflow-gpu PyPIFrom 2.4.02.4.3
tensorflow-gpu PyPIFrom 2.4.02.4.3
tensorflow-gpu PyPIFrom 2.5.02.5.1
tensorflow-gpu PyPIFrom 2.5.02.5.1

Products include Google Tensorflow.

How do I fix CVE-2021-37678?

Upgrade tensorflow to a fixed version for your release line: 2.3.4, 2.4.3 or 2.5.1.

Upgrade, and only deserialise data from trusted sources or with an allow-list of types. More about Unsafe deserialization.

Is CVE-2021-37678 being exploited?

There is no known exploitation of CVE-2021-37678 in the wild and no public exploit code.

Public exploit code: 1 proofs of concept on GitHub.

All Google Tensorflow vulnerabilities

Sources

Also known as BIT-tensorflow-2021-37678, GHSA-r6jx-9g48-2r5r, PYSEC-2021-300, PYSEC-2021-591, PYSEC-2021-789.

Is CVE-2021-37678 in your software?

Upload an SBOM and sbomify checks it against every vulnerability here, every day.

Check your SBOMs for free