Google has announced that two of its newest apps privacy enhancement technology (PET), which includes a type of blurring of objects in video, will be freely available to the public through open source code. New tools are part of Google Computer protected The company says the initiative is designed to transform “how, when and where data is processed to technically ensure data privacy and safety”.
The first was an internal project called Magritte, Now available on Github, which uses machine learning to detect objects and apply a blur effect as soon as they appear on the screen. It can disguise arbitrary objects like license plates, tattoos, etc. “This code is especially useful for video journalists who want to increase privacy,” Google wrote in the blog. “By using this open source, videographers can save time blurring objects from video, while knowing that the underlying ML algorithm can perform detection on video with high accuracy. .”
The other goes by the unwieldy name of “Fully Homomorphic Encryption (FHE) Transpiler, which allows developers to perform computations on encrypted data without being able to access personally identifiable information. Google says it does.” could help industries such as financial services, healthcare and government, “where there is a strong growth in ensuring security around the handling of sensitive data is of the utmost importance.”
Google notes that PET is starting to go mainstream after being primarily an academic exercise. Recent White House technology introduction, says that “it will allow researchers, physicians and others with access to gain insights into sensitive data without having to access the data itself.” Google notes that both the US and UK governments are hold a contest this year to develop PET solutions around financial crime and public health emergencies.
All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission. All prices are correct at time of publication.