Virality of fake news on social media: Are weaponized AI bots to blame, questions Destin Sandlin

A lot of fake news has been spreading in recent times via social media channels such as Facebook, Twitter, Whatsapp, and so on. A group of researchers from the University of Southern California came up with a paper titled “Combating Fake News: A Survey on Identification and Mitigation Techniques” that discusses existing methods and techniques applicable to identification and mitigation of fake news. Microsoft Edge mobile browser also flags untrustworthy news sites with the help of a plugin named NewsGuard. But how far are we in combating the ‘Fake News”?

This weekend, Destin Sandlin, an engineer who conducts an educational video series Smarter Every Day on YouTube, tweeted how Fake News is getting popular on YouTube by being literally engineered within our daily feeds by using sophisticated AI, destructive bots and so on.

https://twitter.com/smartereveryday/status/1091833011262423040

He started off by tweeting about “weaponized bots, algorithm exploitation, countermeasures, and counter-countermeasures!” He mentioned seeing a YouTube video thumbnail with a picture of Donald Trump and Ruth Bader Ginsburg side by side. What caught his eye was, the video received 135,000 views making him feel it’s a legit video. He further explained that the video was simply a bot reading a script. He realized that these bots have come-up with ways to auto-make YouTube videos and upload them. “I recognize that this video is meant to manipulate me so I go to close the video.”

https://twitter.com/smartereveryday/status/1091833831206866944

Sandlin highlighted another fact that these videos had a 2,400 to 143 like to dislike ratio. He believes that this was some sort of weaponized algorithm exploitation.

Source: Twitter

He said that in order to get maximum views on YouTube, all a video has to do is get onto the sidebar or in the suggested videos list. He also mentioned an example of a channel that appeared in his suggestion list, named "The Small Workshop", which managed to get 13 million views.

https://twitter.com/smartereveryday/status/1091835106149453826

The Trump - Ginsburg video

Sandlin searched the YouTube for "After trump sends note to Ginsburg" following which he got tons of different videos but with the same content. He said, “They all use the exact same script, but the computerized voices are different to not trip YouTube's audio detectors, the videos all use different footage to avoid any visual content ID match”. “This is an offensive AI at work, and it's built to avoid every countermeasure”, he added.

Sandlin tweeted, “I think the strategy is simple… if you bot-create enough videos on the same topic and generate traffic to those artificially…many will fail, but eventually, the algorithm will suggest one of them above the others, and it will be promoted as “THE ONE”.”

He further said that tech company engineers are tasked with developing countermeasures to these kinds of attacks. He is dubious of the attacking party he suspects, “Is there a building in a foreign country where soldiers go to work/battle every day to "comment, like, and subscribe?” or are these clever software developers building bots to automatically create videos and accounts to promote those videos? “I would assume they’re using AI to see what types of videos and comments are amplified the most.”

He wonders, “How often Do TTPs (Techniques, Tactics, and Procedures) change?  When the small groups of engineers at YouTube, Facebook, Instagram or Twitter develop a countermeasure, how long until counter-countermeasures are developed and deployed?”

According to a post at Resurgent, “Perhaps Sandlin’s suggestion, responding with an active unity, a countermeasure of forgiveness and grace, is the best answer. There’s no AI or algorithm that can defeat those weapons.”

Read Destin Sandlin’s complete Tweet thread to know more.

WhatsApp limits users to five text forwards to fight against fake news and misinformation

Is Anti-trust regulation coming to Facebook following fake news inquiry made by a global panel in the House of Commons, UK?

Fake news is a danger to democracy. These researchers are using deep learning to model fake news to understand its impact on elections