Galois Wins $1M IARPA Contract To Improve Security Of Data
Galois has been awarded a $1 million contract by The Intelligence Advanced Research Projects Activity (IARPA), in the Office of the Director of National Intelligence, to explore the feasibility of securing data while it is being computed on through the use of fully homomorphic encryption (FHE).
While strides have been made to date in protecting data in transit and data at rest, it has proven far more challenging to protect data while it is being computed on – which is often done between multiple parties or in untrusted environments. Secure computing techniques such as fully homomorphic encryption offer the possibility of general computing on data while it remains encrypted. Practical and easy to use FHE would lead to a sea change in computing. For example, it would enable computation to be outsourced to untrusted computing resources such as cloud-based servers while guaranteeing the privacy of data used in that computation.
For the RApid Machine-learning Processing Applications and Reconfigurable Targeting of Security (RAMPARTS) initiative, Galois, in partnership with the New Jersey Institute of Technology, aims to assess feasibility of a future IARPA program in the area of practical ease of programming, optimizing compilers, and high-performance libraries for fully homomorphic encryption.
“Researchers need to focus on what data analysis needs to be done, rather than how to secure computation when they do that analysis in an untrusted environment, and how to make that secured computation fast,” said Dr. David Archer, Principal Investigator at Galois. “RAMPARTS is designed to prove the feasibility of making fully homomorphic encryption easy to program and optimize in such settings. We and our partners at NJIT look forward to this project opportunity.”
Although FHE is still an emerging technology, its security is well understood, and there has been consistent, substantive, and rapid progress in making it practical from a performance standpoint. However, FHE is still difficult to implement and deploy, and program optimization is still largely unexplored for this technology. RAMPARTS aims to address these pain points so more applications can run efficiently and securely in untrusted environments.