Secure AI Labs (SAIL)


STEX25
Active dates:

May 3, 2019 - May 3, 2019
STEX25
STEX25 Participation:
May 3, 2019 - October 30, 2020
Company information

245 Main St
Cambridge, MA 02142
United States

Elevator Pitch

Elevator Pitch

New data regulations and security threats force companies to silo their data, hurting industry research and innovation. We use a new CPU feature called an "enclave" to encrypt data during its analysis, allowing companies to control how their data is used and by whom, similar to DRM for renting movies. Unfortunately, these enclaves have limited capability, so we built and patented the use of a VM capable of any computation to manage the decentralized control and analysis of remote data silos.
Description

Description

Secure AI Labs is an AI Platform company providing a High-Assurance Federated Learning solution for healthcare. Its products are used by Pharmaceutical and Medical Device companies, Academic Medical Centers, and Research Consortiums to enable Machine Learning and analytics across data sources without aggregating data. Using hardware-based confidential computing platform as its core, SAIL’s defense-in-depth architecture ensures patient data are strongly protected, making it a suitable solution for the enterprise.

Secure AI Labs (SAIL) was founded by researchers and professors from MIT, and is headquartered in Cambridge, MA. Since its founding, SAIL has done successful proofs of concept in bioinformatics and they’re set to start pilots with major hospitals and consulting companies in late summer of this year.

Recent Announcement:
Kidney Cancer Association Uses Secure AI Labs to Fast-track Cancer Research with Federated Learning
https://healthtechhotspot.com/using-federated-learning-to-fast-track-cancer-research/
Technology Description

Technology Description

SAIL’s platform creates a single access point for multiple sources of data (sources focused on one disease as well as sources across complementary endpoints)
Without SAIL’s High-Assurance Federated Learning it would be nearly impossible to manually manage competing data use agreements across so many hospitals, biobanks, and diseases.

To use the data, researchers use SAIL’s SAFE function library to run federated algorithms on the data, and protect it on-prem of the hospital infrastructure.
DRM via AMD’s SEV SNP enclaves automatically manage data authorization, use, and deletion according to data use agreements (translated into digital contracts).
Hardware-assured auditing via enclaves also produces forensic-level auditing for versioning data automatically when patients opt out of studies, notifying IRB’s of change quickly, and recording all for regulatory submission.