# tf-encrypted/tf-encrypted

A Framework for Encrypted Machine Learning in TensorFlow

Repository: https://github.com/tf-encrypted/tf-encrypted
Canonical: https://ross.abutalabs.com/products/tf-encrypted
Homepage: https://tf-encrypted.io/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: secure-computation, machine-learning, tensorflow, privacy, cryptography, deep-learning, confidential-computing
Last push: 2024-09-25T05:32:38+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3087, "days_push": 707, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1243, forks 212 (observed 2026-08-28T04:04:06.666449+00:00)

## What it is
TF Encrypted is a Python framework for privacy-preserving machine learning in TensorFlow, enabling training and prediction on encrypted data using secure multi-party computation and homomorphic encryption. It exposes a Keras-like API so developers can build encrypted ML workflows without cryptography expertise.

## Use cases
- train machine learning models on encrypted data
- run predictions on private data without decrypting it
- build privacy-preserving ML services with secure multi-party computation
- apply homomorphic encryption to TensorFlow models
- protect sensitive data in collaborative ML across parties
- experiment with encrypted deep learning without cryptography expertise

## When to choose
- you need to train or run inference on data that must stay encrypted
- you want privacy-preserving ML with a familiar TensorFlow/Keras-style API
- multiple parties need to jointly compute on private data without sharing it

## When to avoid
- you need cutting-edge performance or the latest MPC research techniques, as the project is in maintenance mode
- you are not using TensorFlow or need framework-agnostic encrypted computation
- you rely on legacy TF1 features like sessions and placeholders

## Facets
- artifact type: framework
- maturity: maintenance
- function: machine-learning, deep-learning, cryptography, privacy, security
- domain: machine-learning, privacy, security, deep-learning
- platform: python, cross-platform
- tags: secure-multi-party-computation, homomorphic-encryption, privacy-preserving-ml, tensorflow, encrypted-computation, confidential-computing

## Member repositories
- tf-encrypted/tf-encrypted (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:06.666449+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:08:21.461229+00:00, confidence not recorded.
  - readme: https://github.com/tf-encrypted/tf-encrypted (fetched 2026-08-28T04:04:06.666449+00:00, sha f0b188406ac2)
  - homepage: https://tf-encrypted.io/ (fetched 2026-08-29T12:20:17.985220+00:00, sha 44136fa355b3)
  - registry_pypi: https://pypi.org/pypi/tf-encrypted/json (fetched 2026-08-29T12:20:17.994986+00:00, sha 3d0af1d67b4c)
- Data as of 2026-08-30T08:39:29.467469+00:00.