github.com
AnonLLM
Anonymize Personally Identifiable Information (PII) for Large Language Model APIs
Independent evaluation guide | 12 synced entries
How to evaluate privacy-preserving ML tools
Reduce sensitive-data exposure and test privacy leakage in ML systems. Compare selection criteria here, then verify every claim against the linked first-party source.
Decision context
What does ml privacy tooling need to cover?
Privacy-preserving machine learning tools reduce or measure the exposure of sensitive information across data preparation, training, evaluation, and inference. Techniques include differential privacy, anonymization, encrypted computation, privacy auditing, and tests for membership inference or model inversion. These controls address different risks: anonymized inputs do not guarantee a trained model cannot memorize records, while a privacy budget does not secure the surrounding data pipeline.
README-synced directory
Privacy-Preserving Machine Learning catalog entries
Each listing preserves the neutral description maintained in the Awesome MLSecOps README and links to the resource's first-party source.
arx.deidentifier.org
ARX - Data Anonymization Tool
Tool for anonymizing datasets
github.com
BMW Anonymization API
This repository allows you to anonymize sensitive information in images/videos. The solution is fully compatible with the DL-based training/inference solutions that we already published/will publish for Object Detection and Semantic Segmentation
veil.ai
Data-Veil
Data masking and anonymization tool
github.com
DeepPrivacy2
A Toolbox for Realistic Image Anonymization
github.com
differential-privacy-library
Library designed for differential privacy and machine learning
github.com
OpenDP
The core library of differential privacy algorithms powering the OpenDP Project
github.com
Privacy Meter
An open-source library to audit data privacy in statistical and machine learning algorithms
github.com
TensorFlow Privacy
Library of privacy-preserving machine learning algorithms and tools
github.com
TF Encrypted
Framework for encrypted deep learning in TensorFlow using secure multi-party computation
github.com
Tool for DATA anonymization
Data anonymization
github.com
Tool for IMG anonymization
Image anonymization
No catalog entries match that search.
Selection framework
How should teams evaluate the options?
Select tools by defining the protected data, attacker access, acceptable utility loss, and compliance evidence you need. For differential privacy, inspect accounting methods, clipping behavior, composition, and how privacy budgets are reported. For anonymization, evaluate re-identification risk using auxiliary-data assumptions that match the threat model, and check whether visual or tabular outputs retain hidden metadata. Privacy attack tools should reproduce realistic access conditions. Favor projects with clear threat assumptions, measurable guarantees, documented limitations, and integration points that keep privacy checks repeatable as models and datasets change.
- 01Explicit privacy guarantees
- 02Utility and accuracy impact
- 03Re-identification testing
- 04Repeatable privacy accounting
Record evidence and limitations for each criterion. A catalog listing is a discovery aid, not a security certification.
Primary references