Facebook has plans to kick off research where it will try to analyze the content of encrypted data without even deciphering it. The social media giant is hiring AI researchers to carry out a study on the matter, reports The Information.
The research could enable Facebook to target ads based on encrypted WhatsApp messages. The company could use the results of the study to encrypt user data without affecting its ad targeting approaches.
Extracting information while keeping data safe
This area of research is known as “homomorphic encryption,” which heavily depends on mathematics. Microsoft, Amazon, and Google also have something similar in works. The purpose behind homomorphic encryption is to enable companies to read and analyze data while keeping it encrypted to protect information from cyberattacks and maintain privacy.
The company told the news outlet that it’s “too early for us to consider homomorphic encryption for WhatsApp at this time.” Facebook could take advantage of the new technology in several ways. Protecting data without reducing the effectiveness of targeted ads will allow the company to achieve business goals and curb concerns of data mismanagement. However, the company could be years away from using homomorphic encryption for its benefit.
Encryption across all platforms
Back in 2019, Facebook revealed its plans to provide end-to-end encryption across all its messaging platforms – Messenger, Instagram, and WhatsApp. The US, Australia, and the UK sent a joint letter urging the social media goliath to not go ahead with the project “without ensuring there will be no reduction in the safety of Facebook users and others, and without providing law enforcement court-authorized access to the content of communications to protect the public, particularly child users.”
Besides, Facebook recently showed off an AI research project called TextStyleBrush that can copy text style in the photo just by looking at one sample word. While existing AI systems could do the same but they require a narrow set of examples. The new tech, however, can recognize a wide range of handwriting and type styles, even with less-than-ideal angles and backgrounds. It’s not perfect to be precise but does the job better than any other AI.
