Secure AI Labs

STEX25 Startup:
May 3, 2019 - October 30, 2020

Data sharing made easy

By: Daniel de Wolff

In theory, sharing and analyzing private data should be a simple equation: data + code = analysis. But in practice, it’s a notorious pain point, and collaborations between companies commonly get hung up in a web of red tape and legal review processes that delay or prevent the discovery of valuable insights. Enter Secure AI Labs (SAIL), the MIT spinout eliminating bottlenecks in the data sharing workflow to help companies create valuable insights from sensitive datasets, like patient records, without compromising data privacy or ownership.

“Our technology solves a huge problem by providing data access to enable research at an accelerated pace,” says SAIL’s CEO and co-founder Anne Kim. “We’re delivering months of gain by cutting through red tape that companies would have to deal with to maintain compliance and preserve privacy. At the same time, our platform allows data scientists to work seamlessly on encrypted data.”

Our technology solves a huge problem by providing data access to enable research at an accelerated pace.

The SAIL platform uses a distributed approach to data analysis called federated machine learning to analyze private data without tedious and dangerous data transfers. SAIL trains algorithm models without having to share or aggregate training data in an unsafe central location commonly called a data lake. "Our platform enables the ability to work with multiple data sets and to decentralize an algorithmic query among the data sets," explains Kim.

But, because sharing code still requires an exchange of proprietary information, the SAIL platform employs secure enclaves to protect training algorithms. These hardware components found in laptops, servers, and mobile devices protect sensitive files like the biometric information on your cell phone. They serve as computing sandboxes within a database to ensure integrity and trust between companies while preventing the theft of intellectual property or the exposure of private information.

Think of biopharma companies attempting to share data with innovative AI startups to generate clinically useful insights from failed clinical trial data. These datasets from sick and dying patients cost more than $1 billion every year to produce and are consistently inaccessible to researchers due to security concerns. The only security solution is a one-way firewall between these companies, creating friction and mistrust when it comes to sharing valuable patient data from clinical trials—which is why SAIL's secure computation technology is such a gamechanger for data sharing.

“The world needs our technology to overcome the standoff that exists between those that have data and those that want to use it,” says Kim’s fellow co-founder Ryan Davis. “We provide a valuable third-party intermediary solution that allows companies to gain insights or train algorithms from data while maintaining privacy and confidentiality. "Companies today sit on an enormous amount of information that provides great value to them, but it can provide even greater value to their partners and collaborators," says Davis. SAIL is currently developing partnerships across the healthcare system to increase and share value from formerly inaccessible datasets.

The world needs our technology to overcome the standoff that exists between those that have data and those that want to use it.

SAIL works with hospitals to help streamline multi-site research studies and industry research collaborations, with pharmaceutical companies trying to fully leverage their clinical trial databases, and with researchers who want to glean valuable insights from siloed or unlinkable databases. And with the help of the MIT Startup Exchange, SAIL is also developing partnerships with customer data analytics teams from insurance, consumer goods, and finance. “STEX25 has been instrumental in allowing us to stretch our legs and work with new industry partners,” says Davis.

The evolution of SAIL is intimately connected to the Institute. CEO Anne Kim's graduate research on clinical trial data sharing forms the framework of SAIL's groundbreaking technology. Meanwhile, Manolis Kellis, a principal investigator of the MIT Computer Science and Artificial Intelligence Lab and head of the Computational Biology group, is a co-founder of SAIL. Additionally, world-renowned data scientist and head of the MIT Connection Science and Human Dynamics labs, Sandy Pentland, is an advisor to the company.

While honing their vision for SAIL's future at MIT's Martin Trust Center and delta v accelerator, Kim and Davis recognized that building a patient data collaboration platform without risking data privacy could accelerate the discovery of new cures and treatments. While healthcare and bioinformatics offers the ideal beachhead for a startup capable of deftly handling private, sensitive, and disparate databases, the SAIL platform is poised to revolutionize data access across a multitude of industries where de-siloing data remains the final frontier for building and deploying new AI solutions. 

Ryan Davis, Cofounder; Anne Kim, Cofounder & CEO, Secure AI Labs (SAIL)