SRI International awarded $8.5 million DARPA contract for data privacy

SRI International of Menlo Park, CA announced February 25 has been awarded an $8.5 million contract by the Defense Advanced Research Projects Agency (DARPA) Brandeis program through the Space and Naval Warfare Systems Command (SPAWAR) to explore how privacy-preserving technologies can be applied to enable information sharing while safeguarding sensitive data across large-scale enterprises.

The vision of the Brandeis program is to break the tension between the need for maintaining privacy and sharing data for collaboration. Researchers will explore the best ways to allow the use of information without losing control of that information. For this contract, SRI will focus on bringing technologies together to enable sharing–and controlling–sensitive data within large, and in many cases multinational, organizations.

“Historically there has been little in the way of privacy safeguards for enterprises wishing to collaborate through extensive data sharing,” said Karen Myers, Ph.D., program director and Principal Scientist in SRI International’s Artificial Intelligence Center. “Our goal for this project is to uncover or develop novel privacy-preserving technologies that strike a balance between unfettered information sharing and maintaining the privacy of information embedded in the data.”

SRI will advance state-of-the-art technologies to facilitate the use of sensitive information for critical applications such as humanitarian relief, while ensuring privacy preservation. SRI will motivate technology development, help define information sharing-policy specification requirements, and provide a platform to deploy, experiment with, and evaluate the effectiveness of these technologies in enterprises. Key technologies within the effort include:

• Modern encryption techniques
• Differential privacy – Focused on the ability to answer queries from large data sets while guaranteeing with high probability that information is not revealed about any individual within the data set.
• Multi-party secure computation – enabling the sharing of encrypted data for use in collaborative computations.

Source: SRI International