Facebook Builds AI Tool to Fool Facial Recognition Systems: Report

Facebook Builds AI Tool to Fool Facial Recognition Systems: Report

The media reported that Facebook's Artificial Intelligence (AI) research team has developed a tool that dodges facial recognition systems to misidentify a person in a video.

 The "de-identification" system, which also works in live video, uses machine learning to change a subject's core facial features in the video, according to a report by VentureBeat on Friday.

"Face recognition can lead to loss of privacy and misuse of face replacement technology to produce misleading videos," reads a paper clarifying the company's approach cited by VentureBeat.

The Verge reported that this de-identification technology had previously worked mostly for still images.

"Advances about recent world events and the misuse of facial recognition techniques require an understanding of the ways that related to de-identification.

 Our contribution is only appropriate for a video, including live video, Is also included, and presents a quality that surpasses the methods of literature., ”Said the paper.

The work is to be presented next week at the International Conference of Computer Vision (ICCV) in Seoul, South Korea.

The development comes at a time when Facebook is undergoing a $ 35 billion (about Rs 2,48,000 crore) class-action lawsuit in Illinois for misuse of facial recognition data. A US court has rejected Facebook's request to quash the lawsuit.

A three-judge panel of Ninth Circuit judges in San Francisco rejected Facebook's petition to dismiss the lawsuit. The case will go on trial until the Supreme Court reports last week.