免费看黄色大片-久久精品毛片-欧美日韩亚洲视频-日韩电影二区-天天射夜夜-色屁屁ts人妖系列二区-欧美色图12p-美女被c出水-日韩的一区二区-美女高潮流白浆视频-日韩精品一区二区久久-全部免费毛片在线播放网站-99精品国产在热久久婷婷-午夜精品理论片-亚洲人成网在线播放

Facebook employs AI technology to remove "revenge porn" content

Source: Xinhua| 2019-03-16 07:51:43|Editor: Shi Yinglun
Video PlayerClose

SAN FRANCISCO, March 15 (Xinhua) -- Facebook said Friday it is using technologies of machine learning and artificial intelligence (AI) to detect and remove non-consensual intimate images, also known as "revenge porn," from its platform.

Antigone Davis, head of global safety at Facebook, said the new technologies can help the company "proactively detect near-nude images or videos that are shared without permission on Facebook and Instagram."

With the AI-based system, such content can be found and sent to human moderators for review "before anyone reports it," Davis said in a blog post.

"When someone's intimate images are shared without their permission it can be devastating," he said.

Davis said that the new technologies were introduced because the victims are often afraid of retribution so they are reluctant to report the content themselves, or are unaware the content has been shared.

He said the new detection mechanism is part of Facebook's pilot program jointly run with victim advocate organizations, which gives Facebook users an emergency option to securely and proactively submit a photo to the company, thus Facebook can create a digital fingerprint of that image and stop it from ever being shared.

Facebook is planning to expand the pilot program in the next few months to benefit more people in an emergency, Davis said.

In order to help people who have been the targets of such destructive exploitation, Facebook also launched a victim-support hub called "Not Without My Consent," where victims can seek support and assistance including measures to have those unauthorized content removed and blocked from further spreading.

In addition, Facebook has offered an appeal process for users to report any posts that are deleted by mistake.

TOP STORIES
EDITOR’S CHOICE
MOST VIEWED
EXPLORE XINHUANET
010020070750000000000000011100001378990931