The two have been working on the project for a year, and according to Google, the open-source privacy infrastructure will help millions of people around the world “develop and launch innovative, differentiated privacy applications that may deliver beneficial insights and services without disclosing any personal data.”
This project has taken the two partners a year to complete. The availability of this privacy infrastructure, according to Google, will aid millions of people in “the global community of developers, researchers, governments, nonprofits, businesses, and others in creating and launching new applications for differential privacy, that can provide useful information and services without revealing information about individuals.”
Since 2019, Google has been working on differential privacy. The online giant’s interest in this field has grown in recent years, driving the creation of this new open-source Python product.
Training third-party specialists to pass on their knowledge to those who wish to learn how to use differential privacy technologies as part of the work with OpenMined.
According to Miguel Guevara, privacy and data product manager at Google, Google approached OpenMined last year to present the concept of building this Python product. The objective was to provide an end-to-end differential privacy solution that was simple and free to use. The researcher says that the project was quickly accepted.
They are particularly thrilled that OpenMined now offers qualified professionals to provide help and resources to any developer wishing to use differential privacy in their projects, in addition to our engineers working together to create and implement the library.
In 2019, Google published an open-source C++, Java, and Go version of their basic differential privacy library. Developers jumped on board immediately, eager to use the library in their products. According to the online giant, startups like Arkhn have utilized it to help hospitals communicate data, while Australian researchers use it for various scientific investigations.
They are also launching a new Differential Privacy Tool, which allows practitioners to view and fine-tune the parameters used to generate differentially private data. Finally, researchers published a summary outlining the strategies they employ to scale differential privacy to datasets of a petabyte or bigger.
The researcher encourages academics and developers to try out the tool and offer comments, noting that Google would “invest in democratizing access to critical privacy-enhancing technology.”