As companies from the financial sector to the health industry amass ever larger, more detailed databases of information about people, it is clear that combining different data sets can offer powerful insights. But to protect users’ privacy, many of these data sets stay locked up inside corporate firewalls.
Chipmaker Intel thinks it has a way to let valuable data be combined and analyzed without endangering anyone’s privacy. Its researchers are testing a super-secure data locker where a company could combine its sensitive data with that from another party without either side risking that raw information being seen or stolen.
The technology could allow companies to do much more with their data, to their benefit and the benefit of their customers. Today, companies either keep their data closely held, or release it with certain details obscured or removed, which limits how useful it can be. Health-related data is particularly limited today, due to HIPAA regulations that govern the sharing of patient records. Two parties using the system need to agree on the analysis to be performed, feed in their data, and then get the results without either side seeing the other’s raw input. The original information is then wiped from the locker. “There are many companies and organizations that own data, and would like to share in a way that [ensures] it is not released or stolen,” says Sridhar Iyengar, a director of security research at Intel Labs. “This is a neutral environment where parties can place their data and derive an answer without revealing their data to one another.”
The project, known as Reliance Point, is a collaboration between Iyengar’s research group and Intel’s data center group. When the Reliance Point system boots up, a security chip is used to check that the BIOS, the lowest-level software on a computer that starts it up, hasn’t been tampered with. The BIOS then makes its own checks before activating the next level of software, which in turn makes its own checks, a chain-like process that continues until the system is fully operational.
All those checks generate data that parties using Reliance Point can use to assure themselves the system can be trusted before they feed in their precious data. “They can have high confidence that the platform has not been tampered with,” says Iyengar.
Gerome Miklau, an associate professor at the University of Massachusetts, Amherst, says that the approach Intel is testing does have the potential to reduce risks of disclosing sensitive information. But he adds that it could be difficult to a find third party that companies would trust to run such a service.
Agreeing on what code is to be run on combined data might also be difficult to negotiate, says Miklau, whose own research examines how to manage and use large-scale data sets without harming privacy. “Verifying that code does what you think it does is always hard,” he adds. “There is also the problem of deciding what the output of the code reveals about the input and whether that disclosure is acceptable.”
Intel is currently working on a larger version of the prototype system, and plans to test it using some real data sets. Nikhil Deshpandea, a senior business strategist at Intel Labs, says doing that should demonstrate what benefits the technology might offer to companies. “There’s a whole bunch of data stuck in these silos because they cannot talk to one another.”
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Everything you need to know about artificial wombs
Artificial wombs are nearing human trials. But the goal is to save the littlest preemies, not replace the uterus.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.