A group of computer scientists shocked the media last month when they proved, yet again, the inadequacy of de-identification as a means to protect the privacy of shared data. The findings came as little surprise to experts in privacy technologies. This pressing need to protect privacy, including against re-identification, motivates us at Duality to provide innovative technologies to protect data privacy in the age of AI and Machine Learning.
Duality brings to the market a privacy-protected data science platform optimized for Homomorphic Encryption (HE). Often referred to as the “Holy Grail of Data Privacy”, HE enables computations on encrypted data and secure collaboration among multiple parties. Having matured in the academic realm during the early part of the past decade, HE is now being productized, to respond to the growing market needs for secure cloud computing and privacy-preserving AI capabilities.
We at Duality Technologies are excited to be part of the Homomorphic Encryption Standards consortium, collaborating with Intel, Microsoft Research and additional organizations on the creation of a privacy-preserving HE computation ecosystem. Industry standards ease adoption of the technology by even the most conservative users, and foster increased market participation on all layers of the secure computing stack, from optimized hardware and platforms all the way to applications and solutions for a broad range of use cases.
A handful of HE implementations exist, including the PALISADE open-source library on which Duality’s data science platform is built. PALISADE development was led by our CTO, Dr. Kurt Rohloff, who coordinated the efforts of multiple teams of privacy professionals, defense contractors and academics, based on insights from two earlier generations of HE implementations. Designed with a focus on practical applications of encrypted computing, PALISADE offers users rapid configuration for run-time optimization and modular design to support advanced analytics and data science projects.
Establishing a standard is crucial to uniformize and simplify APIs across libraries, including industry-leading libraries such as PALISADE. This in return makes it easier for application developers to adopt HE and integrate it into their analytics and collaboration solutions. Furthermore, in a standardized HE ecosystem, applications will be portable, offering users a maximum of flexibility and business value.
It is doubtless that standards are accelerating privacy-enhanced information sharing across regulated industries. With new data privacy regulations emerging worldwide, businesses, research organizations, and regulators are currently looking for technologies to leverage data analytics and AI in a compliant manner. Such technologies, for example, are required to protect the privacy of patients who participate in large-scale genomic studies. In the financial industry, multiple banks can collaborate on their joint data to identify financial criminals and money launderers without exposing PII.
At this age of Big Data, AI and digital distrust, use cases of HE are almost endless. Establishing an industry standard is the next step towards a privacy-enhanced future.