CVE-2022-23561: Out of bounds write in TFLite
Impact An attacker can craft a TFLite model that would cause a write outside of bounds of an array in TFLite. In fact, the attacker can override the linked list used by the memory allocator. This can be leveraged for an arbitrary write primitive under certain conditions.
Patches We have patched the issue in GitHub commit 6c0b2b70eeee588591680f5b7d5d38175fd7cdf6. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range. 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 Wang Xuan of Qihoo 360 AIVul Team.
Other sources
Tensorflow is an Open Source Machine Learning Framework. An attacker can craft a TFLite model that would cause a write outside of bounds of an array in TFLite. In fact, the attacker can override the linked list used by the memory allocator. This can be leveraged for an arbitrary write primitive under certain conditions. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2022-23561?
CVE-2022-23561 is classified as a high severity vulnerability due to its potential for arbitrary writes.
How do I fix CVE-2022-23561?
To fix CVE-2022-23561, upgrade to TensorFlow version 2.7.1 or later.
Which versions of TensorFlow are affected by CVE-2022-23561?
CVE-2022-23561 affects TensorFlow versions up to 2.5.2, 2.6.0 to 2.6.2, and 2.7.0.
What type of vulnerability is CVE-2022-23561?
CVE-2022-23561 is a write outside of bounds vulnerability specific to TFLite models.
Can CVE-2022-23561 be exploited remotely?
Yes, CVE-2022-23561 can potentially be exploited remotely if an attacker crafts a malicious TFLite model.