Google Open Sources Privacy-Focused Blur Tool Magritte

Dec. 23 (Portaltic/EP) –

Google ha shared with open source the magritte projecta privacy enhancement technology (PET for its acronym in English), which blur objects in a video for data protectiona novelty that has been announced together with the performance improvements of Fully Homomorphic Encryption (FHE), another solution dedicated to performing calculations on encrypted data without accessing the information.

PETs are technologies with which Google searches provide “great experiences” with all its functions, always keeping user information protected and ensuring their anonymity.

However, the tech giant has indicated that it is aware that these technologies are expensive and have many entry barriers for its implementation in the industries and that, in addition, require large computational resources, as explained on his blog.

In this sense, it has advanced that it seeks to create a safer ecosystem on the Internet and has focused on democratizing the access of PETs. Thus, one of his technologies, called Magritte, will become an open source project. Thanks to this, any user will have access to it and You will be able to improve privacy features.

Specifically, Magritte is an internal Google project and, thanks to this solution, you can blur objects like license plates, posters, tattoos in images and videos. For it, uses machine learning (‘machine learning’), in order to detect objects and apply an automatic blur “as soon as they appear on the screen.”

With this tool, tasks such as those of video journalists they are benefited with greater guarantees of privacy. In this way, information professionals will be able to save time in manually blurring certain objects in a video, since the algorithm is capable of detecting and hiding it.

IMPROVEMENTS IN THE FHE TRANSPILATOR

On the other hand, the American multinational has announced that it has rMade performance improvements for the Fully Homomorphic Encryption compiler (FHE), which he already launched publicly last year.

Specifically, a transpiler is a special compiler that converts the source code of a programming language into another equivalent. In this framework, FHE is a Technology that allows developers to perform calculations on encrypted data without accessing personally identifiable information.

With the improvements implemented, Google has included new circuit optimizations for the transpiler, so it has reduced both its cost and computational time, which has meant the execution of two of his “biggest challenges” of , as he has pointed out in this publication. Thus, the size of the circuit has decreased by 50 percentimproving overall speed.

This technology is Useful in financial services industries, healthcare, and even governmentsbecause it offers a “strong” guarantee that data processing is secure and confidential.

Finally, Google has stressed that organizations and governments “around the world” are also exploring the use of PET to face social challenges. So much so, that the United States and United Kingdom Government they organized a contest this year for develop PET solutions regarding financial crime and public health emergencies.

“We are excited about the new tests and feedback on our open source PETs and look forward to release more updates in 2023 and beyond”, has concluded the technological manufacturer.