Facebook rolls out photo/video fact checking so partners can train its AI

Sometimes the fake news lives inside Facebook as photos and videos designed to drive disinformation campaigns, instead of posting off-site news articles that can generate their own advertising revenue. To combat these meddlesome with political rather than financial motivation, Facebook has to be able to detect false news within the images and the audio that accompanies the video clips. Today it is expanding its verification program of photo and video events from four countries to its 23 fact-checking partners in 17 countries.

"Many of our external fact-checking partners have experience in evaluating photos and videos and are trained in visual verification techniques, such as inverse image search and image metadata analysis, such as when and where the photo or photo was taken. video, "says Facebook product manager Antonia Woodford. "As we get more ratings from data inspectors in photos and videos, we can improve the accuracy of our machine learning model"

The goal is for Facebook to automatically detect manipulated images, out of context. images that do not show what they say they do, or text and audio statements that are provably false.

In the epic security manifesto of 3,260 words of the previous night, Facebook CEO Mark Zuckerberg explained that "the definition of success is that we stop". cyber attacks and coordinated information operations before they can cause damage. "That means using AI to proactively search for fake news instead of waiting for users to mark them." For that, Facebook needs AI training information that will be produced as an escape from its partners' photo and video verification operations. 19659003] Facebook is developing technological tools to assist its inspectors with facts in this process. " we use optical character recognition (OCR) to extract text from photos and compare that text with article holders of inspectors of acts. We are also working on new ways to detect if a photo or video has been tampered with, "says Woodford, referring to DeepFakes using AI video editing software to make someone appear to say or do something they do not have.

Image memes were one of the most popular forms of misinformation used by the interceptors of the IRA elections in Russia.The problem is that, since they are easily shareable and do not require people to leave Facebook to see them , can get a viral distribution of unsuspecting users who do not realize they have become pawns in a disinformation campaign 19659003] Facebook could potentially use the high level of technical resources needed to build the AI ​​for counterfeit news with detection of memes as an argument as to why Facebook should not be divided with Facebook, Messenger, Instagram and WhatsApp combined , the company gains economies of scale when it comes to fighting the scourge of misinformation.