Cart ()

Can Facebook Enforce Their Ban On Deepfakes Ahead of the 2020 Race?

Can Facebook Enforce Their Ban On Deepfakes Ahead of the 2020 Race?

As the United States prepares for another divisive presidential election, social media institutions are increasingly facing calls to help curb the spread of online political propaganda. Recently, Facebook announced a blanket ban on manipulated videos and photos using “DeepFake” technology, but importantly, this does not extend to all false content on the platform. Understandably, questions have emerged regarding whether Facebook can even enforce such a ban without false positives and mass censorship. The company’s answers won’t be smoothing tensions anytime soon.

It was originally leaked to The Washington Post that Facebook had already begun banning users from posting Deepfakes, classified as a piece of video or audio material where highly convincing editing software has manipulated how a subject looks, sounds and acts.

“While these videos are still rare on the Internet, they present a significant challenge for our industry and society as their use increases,” said Monika Bickert, the company’s vice president for global policy management, in a recent blog post. I would go a step further to suggest it’s an existential threat to our modern information age.

In the announcement, Facebook tries to play the slippery centrist in declaring it “does not prohibit all doctored videos”, just those considered to be Deepfakes, meaning they must be false material “edited in ways that aren’t apparent to an average person and would likely mislead someone” and must use artificial intelligence software to achieve this goal. The only distinction here between a Deepfake and all doctored videos is methods, not results, which begs questions about how Facebook can tell the difference and whether these differences constitute meaningful distinctions.

A prime example of this false debate involves a viral video of Democratic House Speaker Nancy Pelosi, in which she was edited to be slowed down, appearing to slur her words as though she was drunk, which was viewed by millions across the platform. Political camps argued not on the merits of its falsehood, but rather the means by which the falsehood was created, each taking a side on whether it was a Deepfake or something made without the use of artificial intelligence software—what disinformation researchers are now calling “cheapfakes” or “shallowfakes.” The reason this should be considered a false debate is that the distinction has no substantial difference; the result was effectively the same, whether conducted by humans or robots.

As recently acknowledged by The Verge, Facebook’s policy change came just a few days before a scheduled hearing on manipulated media by the House of Energy and Commerce. Bickert spoke on Facebook’s behalf during the hearing, reiterating numerous times the company’s promise to allow users to have control over their own data and feed, yet avoided direct questions on how it would enforce the Deepfake ban, how it would determine a potential Deepfake’s creation methods, and other pressing issues.

Facebook’s new policy — now known as Enforcing Against Manipulated Media (EAMM) — was immediately chastised by Subcommittee Chairwoman Rep. Jan Schakowsky (D-IL) as a product of “Congress, unfortunately, taking a laissez-faire approach to regulating unfair and deceptive practices online over the past decade”, questioning why “big tech [that’s] failed to respond to the grave threat posed by deep-fakes” can regulate itself at all. 

Unfortunately, Schakowsky’s choice to use the Pelosi video as an example of such a “Deepfake threat” during the hearing shows that the public sector is also sorely lacking in its capacity to be a competent big tech watchdog. 

Nevertheless, the hearing wasn’t all semantic floundering. Schakowsky was quick to cite Section 230 of the Communications Decency Act, “which provided online platforms like Facebook a legal liability shield for 3rd party content. Many have argued that this liability shield resulted in online platforms not adequately policing their platforms, including online piracy and extremist content. Thus, here we are, with big tech wholly unprepared to tackle the challenges we face today.” This pivotal law, which has been a topic of controversy for several years, was declared “a topline concern for this subcommittee,” directly suggesting that new policy reforms will come down the line.

“We must protect consumers regardless of whether they are online or not” Schakowsky declared. “For too long, big tech has argued that ecommerce and digital platforms deserved special treatment and a light regulatory touch. Federal Trade Commission works to protect Americans from many unfair and deceptive practices, but a lack of resources, authority, and even a lack of will has left many American consumers feeling helpless in the digital world. Adding to that feeling of helplessness, new technologies are increasing the scope and scale of the problem. Deepfakes, manipulated video, dark patterns, bots, and other technologies are hurting us in direct and indirect ways.”

Bickert’s response, on the other hand, provides little reassurance concerning how Facebook plans to remain the biggest platform in the world —home to over 2.3 billion users and countless piles of data uploaded by the minute — while actively administering this new policy with any sense of clarity, especially if Deepfakes are an existential threat to the online information sphere. 

“People share millions of photos and videos on Facebook every day, creating some of the most compelling and creative visuals on our platform,” she stated. “Some of that content is manipulated, often for benign reasons, like making a video sharper or audio more clear. But there are people who engage in media manipulation in order to mislead, and these manipulations can be made through simple technology like Photoshop or through sophisticated tools that use artificial intelligence or ‘deep learning’. While these videos are still rare on the Internet, they [nevertheless] present a significant challenge for our industry and society as their use increases.”

For future hearings to be worthwhile, misinformation at large must be made a clear and key topic of interest, an opinion that also seems to be shared by House Intelligence Committee Chairman Rep. Adam Schiff (D-CA), who cautioned the government against falling for these false distinction traps. 

“As with any new policy,” Schiff stated, “it will be vital to see how it is implemented, and particularly whether Facebook can effectively detect deepfakes at the speed and scale required to prevent them from going viral,” emphasizing that “the damage done by a convincing deepfake, or a cruder piece of misinformation, is long-lasting, and not undone when the deception is exposed, making speedy takedowns the utmost priority. We’ll also be focused on how Facebook deals with other harmful disinformation like so-called ‘cheapfakes,’ which are not covered by this new policy because they are created with less sophisticated techniques but nonetheless purposefully and maliciously distort an existing piece of media.”

Adding to the confusion, Facebook spokesman Andy Stone recently told The Washington Post that the ban on such manipulated-media “would not apply to political advertising”, noting the company has also declined to send such political ads to their own third-party fact-checkers or even label the ads as being false. 

So what does the ban actually cover in a political context? Any material that wasn’t accompanied by a cheque to Facebook? Do these conflicts of interest not affect what the company decides is and is not a Deepfake? Or are we expected to simply ignore the problem, as Facebook has so brazenly suggested with its new opt-out feature for political advertising? 

The creator of the Pelosi video also told The Post last year his video was a form of satire and “a cautionary tale of technology and democracy.” Given the policy’s provisions don’t actually cover parody or satire, we could be on the edge of another contentious debate over what is considered “deceptive” versus dramatic, adding yet another layer of complexity to this already mired issue.