In order to ensure a certain level of privacy protection for Internet users, Google is developing a series of tools called “privacy-enhancing technologies” or PET. And the firm has been sharing some of these tools for a few years now, making them open source for the community.
This December, Google is releasing new open-source privacy software. The technology, already used by Google engineers, is called Magritte.
According to Google, Magritte uses machine learning (a form of artificial intelligence) to detect objects while using few hardware resources. Then, the software can automatically blur these objects as soon as they appear on the screen.
“This code is particularly useful for video journalists who want to provide increased privacy guarantees. By using this open source code, videographers can save time blurring objects in a video, knowing that the underlying ML algorithm can perform detection on a video with high accuracy”explains the Mountain View firm.
Magritte, the new open source software from Google
But of course, since the technology is now open source, it could also be used by other companies, including Google’s competitors. Moreover, in its announcement, the Mountain View company explains that while its PET (privacy-enhancing technologies) already improve the protection of user data on Google products, the adoption of this type of technology by the whole industry can be complicated. “[…] due to many barriers to entry, including requiring large computational resources, being complex to manage, and expensive to implement”adds the firm.
As a reminder, it was in 2019 that Google began to open source technologies that it believes can bring important improvements in terms of privacy protection. The firm then shared a software library for a technology called “differential privacy”.
“Differential analysis of private data is a principles-based approach that allows organizations to learn from the majority of their data while simultaneously ensuring that those findings do not distinguish or re-identify an individual’s data” , explained Google.
“For example, if you’re a health researcher, you might want to compare the average length of stay for patients at various hospitals to see if there are any differences in care. Differential privacy is a high-assurance, analytical way to ensure use cases like this are handled in a privacy-preserving way.”had also indicated Google.
