CVE-2020-5215: TensorFlow Denial of Service Vulnerability
Summary
CVE-2020-5215: In TensorFlow before 1.15.2 and 2.0.1, converting a string (from Python) to a tf.float16 value results in a segmentation fault in eager mode as the format checks for this use case are only in the graph mode. Affects tensorflow across 3 version ranges.
< 1.15.2
>= 2.0.0, < 2.0.1
< 5ac1b9e24ff6afc465756edf845d2e9660bd34bf
Fixed in 1.15.2 or 2.0.1, depending on the release line you run. How to fix ↓
Also tracked for
- tensorflow · Bitnami
- tensorflow-cpu · PyPI
- tensorflow-gpu · PyPI
The same CVE-2020-5215 is tracked separately for each ecosystem and vendor. Its fix information is the same; the package and affected versions differ.
Check if this vulnerability affects your application
Upload a manifest or SBOM and we'll show whether you're affected, plus every other risky package in the same scan.
No signup required · Results in about 60 seconds
Background
TensorFlow is an open-source machine learning framework developed by Google. It provides a comprehensive ecosystem of tools, libraries, and resources for building and deploying machine learning models. TensorFlow supports both graph mode and eager mode execution. In graph mode, computations are defined as a static graph, while in eager mode, computations are executed immediately. TensorFlow allows conversion of data types, including converting strings to tf.float16 values.
Vulnerability Detail
The vulnerability (CVE-2020-5215) exists in TensorFlow versions before 1.15.2 and 2.0.1. When converting a string to a tf.float16 value in eager mode, a segmentation fault occurs. This issue is due to the lack of format checks for this use case in eager mode, as they are only present in graph mode. An attacker can exploit this vulnerability by sending a data point containing a string instead of a tf.float16 value, leading to a denial of service in inference/training. Additionally, manipulating saved models and checkpoints by replacing a scalar tf.float16 value with a scalar string can trigger this issue. The vulnerability can be easily reproduced by using the tf.constant("hello", tf.float16) function with eager execution enabled.
How To Fix
Latest published version: 2.21.0 as of 22 Oct 2026
pip install --upgrade "tensorflow==2.21.0"Upgrading to it may include breaking changes; the versions below are the minimum fixed release for this vulnerability.
If you can't move to the latest
The minimum fixed version depends on the release line you run.
-
< 1.15.21.15.2pip install --upgrade "tensorflow==1.15.2" -
>= 2.0.0, < 2.0.12.0.1pip install --upgrade "tensorflow==2.0.1"
Tip — Not sure which release line you're on? Vulert can find the fix for the version you are using. Scan free →
Workaround
There is no specific workaround mentioned for this vulnerability.
Conclusion
The vulnerability (CVE-2020-5215) in TensorFlow allows for a denial of service attack in inference/training. It can be exploited by sending a string instead of a tf.float16 value or manipulating saved models and checkpoints. Upgrading to TensorFlow 1.15.1, 2.0.1, or 2.1.0 resolves the issue. For further support on vulnerability remediation, contact DevNack.com.
See every vulnerability recorded for tensorflow →
FAQ
TensorFlow is an open-source machine learning framework developed by Google. It provides tools and resources for building and deploying machine learning models.
The vulnerability can lead to denial of service in inference/training in TensorFlow. An attacker can send a data point containing a string instead of a tf.float16 value, causing a segmentation fault.
The vulnerability can be exploited by sending a data point with a string instead of a tf.float16 value or by manipulating saved models and checkpoints to replace a tf.float16 value with a string.
TensorFlow versions before 1.15.2 and 2.0.1 are affected by this vulnerability.
To fix the vulnerability, upgrade to TensorFlow 1.15.1 or 2.0.1, or switch to TensorFlow 2.1.0, which is not affected.
References
Related vulnerabilities
Recent vulnerabilities recorded for tensorflow.
