Data analysis and marketing vendor AppsFlyer announced a long-term partnership with Intel earlier this month in hopes of accelerating the commercial viability of homomorphic encryption.
For those that aren’t familiar, fully homomorphic encryption (FHE) is hailed as the “holy grail” of confidential computing. Compared to traditional confidential computing environments where sensitive data is decrypted and processed within a secure enclave on the server, homomorphic encryption offers several advantages, particularly where privacy is concerned.
Perhaps the biggest promise of technology is the ability to process data without decrypting it first. However, homomorphic encryption still in its infancy. Early implementations have massive caveats when it comes to performance, even for simple operations, according to Intel Labs engineer Rosario Cammarota.
The Future of the Data EconomyFor a company like AppsFlyer, homomorphic encryption is the future of online marketing.
The company specializes in mobile marketing analytics and offers a host of tools designed to help customers and partners make more informed decisions based on data acquired by their users.
“You can think about it as a huge marketplace where partners and customers share data and work together to better achieve insights and to run better campaigns,” Edik Mitelman, GM of AppsFlyer’s marketing operating system team, told SDxCentral.
“Ever since the internet was created, the entire ecosystem was built on user-level data,” he explained. “We give our user data to companies, and we expect a great user experience in return.”
However, over the last few years, privacy concerns have forced enterprises to reassess how they collect, store, and commercialize user data, Mitelman explained.
The vendor’s Privacy Cloud, which launched alongside the company’s partnership with Intel, attempts to thread the needle between collecting usable data and protecting user’s privacy.
The platform is a data clean room. Companies enter their user-level data into the Privacy Cloud where it's stripped of identifiers, analyzed, and aggregated. Only then are the results shared with the appropriate parties.
The approach, Mitelman claims, drastically improves user privacy because it limits the amount of data changing hands to draw these insights.
Privacy's New FrontierImplementing homomorphic encryption is the next step to ensuring user data stays private, Mitelman explained. However, making the jump won't be easy, especially at the company's scale.
“Homomorphic encryption is not ready yet for a scale of data of a marketing ecosystem,” he said. “AppsFlyer processes more than 120 billion events per day. Try to encrypt that scale of data, you will fail.”
That’s where the collaboration with Intel comes in. Homomorphic encryption is one of several security technologies under development at Intel. And while still in the proof-of-concept phase, homomorphic encryption promises to enable end-to-end data privacy.
“Creating zero-trust environments for data at our scale has never been done before,” Mitelman said. “When it’s ready, it will bring the privacy-preserving technologies to the next frontier.”
DARPA Taps Intel for Homomorphic AccelerationThe collaboration with AppsFlyer is far from Intel’s first in this space. Earlier this year, the Defense Advanced Research Projects Agency (DARPA) tapped the chipmaker to develop accelerators for fully homomorphic encryption applications.
The chips will eventually be tested and deployed in Microsoft Azure to secure government workloads with the ultimate goal of enabling sensitive government data that can be stored, transported, and processed without decrypting it at any point.
However doing so requires circumventing performance bottlenecks, which Intel plans to address using a custom ASICs currently in development.
“When fully realized, the accelerator could deliver a massive improvement in executing FHE workloads over existing CPU-driven systems, potentially reducing the time to processing the cryptograms by five orders of magnitude,” the company claims.
Comments