The company says the new tools are part of Google’s Protected Computing initiative, designed to transform “how, when and where data is processed to technically ensure its privacy and security.”
The first is an internal project called Magritte, now available on Github, that uses machine learning to detect objects and apply a blur as they appear on the screen. You can hide plates, tattoos and similar random objects with it. “This code is particularly useful for video reporters who want to be more assured of privacy,” Google wrote on the blog. “Using this open source code, videographers can save time blurring objects in a video.”
Google launches the highly anticipated feature!
The other code, long called “Fully Homomorphic Encryption (FHE) Transpiler,” allows developers to perform calculations on encrypted data without accessing personally identifiable information. Google says this could help industries such as financial services, healthcare, and government: “where the security assurance regarding the handling of sensitive data is paramount.”
Google notes that PETs are starting to enter the mainstream after a mostly academic exercise. The White House recently launched the technology, saying it will “allow researchers, doctors, and others with permission to gain insights from sensitive data without access to the full range of data.” Google noted that both the US and UK governments are holding a competition this year to develop PET solutions around financial crimes and public health emergencies.