End-to-end encrypted ML inference with Amazon SageMaker AI and FHE
Amazon just made encrypted ML inference practical. No more choosing between privacy and speed.
Why it matters
AWS is shipping a higher-level, production-ready path for FHE-based inference on SageMaker, moving encrypted ML from academic exercise to deployable infrastructure. This matters for regulated industries (healthcare, finance) that need real-time inference without exposing raw data.
The key facts
10 to knowAmazon SageMaker now supports end-to-end encrypted ML inference via FHE
concrete-ml library enables scikit-learn API compatibility for FHE models
Approach replaces low-level hand-crafted algorithms (SEAL library) with high-level abstraction
Supports 'several common types of models out of the box'
Published June 2026
Amazon SageMaker now supports concrete-ml for FHE-based inference
concrete-ml is scikit-learn API compatible
Supports multiple model types out-of-the-box vs. hand-crafted linear regression
Enables end-to-end encrypted inference without low-level SEAL library coding
Published June 2026 on AWS ML blog
Go to the source
AWS Machine Learning Blogaws.amazon.com
Publisher excerpt: This blog has previously discussed FHE for ML inference in the post Enable fully homomorphic encryption with Amazon SageMaker endpoints for secure, real-time inferencing, but this post goes a little further. That previous post showed how to implement FHE-based inference 'from scratch' by…