Hell froze over.
Kudos to whomever at Google is responsible for google’s decision, revealed today, to enable differential privacy for its tensor flow machine learning service.
Most of us first heard of differential privacy when Apple, almost three years ago, introduced it as the technology that would enable Apple to intelligently learn about and help you organize your data with them (think: “Siri show me pictures of trees”), without being able to personally identify that you take a lot of pictures of trees. (Googles smart reply feature in gmail, in stark contrast, reads and personally associates every word YOU have written in email to provide its suggestions).
With Google’s Tensor Flow service now offering Differential Privacy, developers could build a service to reply to emails, sort photos, or drive you, without the need to personally identify you with the data required to accomplish this task.
The outstanding questions then, are, will differential privacy spread throughout Google and make Google into a new different company that actually values and protects users privacy like DuckDuckGo? Will it destabilize google core business? Will it highlight googles opposing data collecting behavior to users? or will it enable a new company to rise and display Google by using Google’s own technology?
No matter the results, the news that Google has followed Apple into Differential Privacy is a victory for regular computer users all over the world. Privacy is a fundamental human right.