Facebook works on reading users’ encrypted messages

Homomorphic Encryption in practice will be used in the Facebook ads network

A team of Facebook artificial intelligence researchers is reportedly working on analyzing the content of users’ encrypted messages without having to decrypt them.

The project is based on what is known as “homomorphic encryption”, a form of encryption that allows computations on encrypted data without first decrypting it. This technology will enable Facebook to analyze the content of the messages exchanged by users, without, however, reading the text itself. In fact, “homomorphic encryption” is not limited to messages, but can be extended to any encrypted data.

Companies such as Google, Amazon, and Microsoft are also working on “homomorphic encryption” in their attempt to gather users’ information. This technology allows them – as well as third-party companies – to perform analysis of encrypted information without uncovering the raw data.

Facebook plans to deploy the “homomorphic encryption” model through Messenger and Instagram direct messages and gather the information that will be used to the company’s ads network. Both users’ messages on Instagram and Messenger are currently encrypted, but Facebook could use this technology and obtain personal data while also maintaining a level of privacy.

Homomorphic Encryption in practice

During a recent interview, IBM’s Flavio Bergamaschi explained what is “homomorphic encryption” and how it can be used in practice. IBM is among the companies which are already working on this new technology and exploring its use in areas such as health and finance.

“Now you can outsource computation without giving access to the data. You allow other entities to perform the computation for you or share data with other people without giving access to your raw data” said Bergamaschi.

“When you are just walking down the street and you want to find a coffee shop, today, you just pick up your phone and ask your phone where there is a coffee shop. But when you do that you revealing to the entity that is about to tell you where there is a coffee shop, your location, the time of that question, where you are, and that you want a coffee shop”. According to Bergamaschi if this process is performing through the use of “homomorphic encryption”, the entity completing the computation is not able to decrypt raw data, such as user’s location, but it has the ability to perform the match and return the information.

Targeted Advertisements

According to analysts, Facebook will use this technology to enhance the targeted ads it displays to users. In practice, the company will be able to decrypt users’ conversations, maintaining encryption, and utilizing data, with experts estimating that this will bring a lot of revenue to Facebook’s coffers.

At the moment, no more information has been leaked about how Facebook plans to use “homomorphic encryption”, while there were reports that this technology will be applied to WhatsApp messages as well.

However, Will Cathcart, head of WhatsApp hurried to point out that this is not the case at present. “We are not pursuing homomorphic encryption for WhatsApp. I have been asked this before. We should be skeptical of technical claims that apps like ours could see messages in “good” cases only. That’s just not how technology works” he tweeted.

The team of Facebook AI experts continues to work on the “homomorphic encryption” project, although there is no timetable for when Facebook will announce results if it does.

George Mavridis is a freelance journalist and writer based in Greece. His work primarily covers tech, innovation, social media, digital communication, and politics. He graduated from the Aristotle University of Thessaloniki with a BA in Journalism and Mass Communication. Also, he holds an MA in Media and Communication Studies from the Malmö University of Sweden and an MA in Digital Humanities from the Linnaeus University of Sweden.