Authorities must tackle deepfakes before the train has left the station

The world has already learned the hard way that digital media - and especially social media - can weaken societies and undermine our systems of government. We are repeating the same mistakes with deepfake technology, as it becomes more readily available, cheaper, and better. 

We must act now to create effective laws and policies against deepfakes before they are used against us - just as social media was previously. The threat posed, which allows users to create highly realistic, doctored video content of individuals by (for example) superimposing their face onto another’s body, might well be impossible to quantify. But it is significant.

Some deepfake technology is currently used for harmless entertainment, art or more dangerously, pornographically, to target women. One study found that 96 percent of all deepfake videos online targeted women. However it has another dark side; it can be used by foreign or domestic actors to create the ultimate in ‘fake news’ and influence behaviour and elections. 

Researchers at an MIT conference were able to create a real-time interview with a ‘Vladimir Putin’. Outside the ‘safe space’ of an academic conference, its applications are endless - and terrifying. It has the potential to accelerate our trajectory towards a post-truth era that neither our populations nor institutions are prepared for.

Our governments are already struggling to adapt to the shift from ‘few-to-many’, to ‘many-to-many’ content distribution. Deepfake video content raises the stakes since it is much more psychologically flammable than static images. We are more receptive to video and audio content, in a way that we no longer are to photos. This makes deepfake video more dangerous than any other type of misinformation.

The consequences of not getting a hold on this are terrifying. The deepfake universe could negate almost everything we see on a screen. We could take one look at our screen showing a video of a terrorist chemical attack just down the road, and completely ignore it in the belief that it’s another fake.

Or fake videos could do the same thing to everything in our reality - perhaps an effective weapon for any dictatorship’s propaganda machine as he tries to silence dissent. ‘Leaked’ videos could be used to justify foreign wars. 

And all of this is before the bots get involved. Facebook currently predicts that approximately 6 million bots are infesting its platform; it was these bots who were responsible for posting a large portion of political content in 2016. Once bots can create their own deepfakes, the problem becomes exponential.

In a September 2019 study, an Amsterdam-based company named Deeptrace found 14,678 deepfake videos on popular streaming websites - double the number from December 2018. As worrying, there are no specific national laws against deepfakes in the US or UK. We can get ‘Boris’ to announce a blockade of Ireland tomorrow at 5am. Or have the Federal Reserve raise interest rates.

Misinformation is hard for government to police, despite increased resources during the pandemic. I would still encourage that deepfakes be treated as criminal activity because of the huge harm they can cause. 

Texas passed a law in September, criminalising the publishing and distribution of deepfake videos intended to harm a candidate or influence results within 30 days of an election. Other states like California have implemented similar laws aimed at election interference. 

We need national (ideally transnational) laws, backed with serious enforcement. Since many of the content creators are likely to be based overseas, possibly in countries without extradition treaties, law enforcement alone will never be enough.

Criminal charges against anyone sharing content would risk criminalising millions of individuals who (perhaps innocently) shared what they thought was real. This is where the analogy with child pornography breaks down: whereas it is difficult to imagine someone genuinely not knowing that they were sharing child pornography, the very nature of deepfakes means that most sharers could be genuinely duped. 

Big Tech must therefore help identify deepfakes, and swiftly eliminate them. Those firms have the capacity to block overtly illegal content, like child or 'revenge' pornography for example. 

Governments and the private sector need to invest in applications of technologies like AI and Blockchain that can help to certify genuine content. There will be inevitable objections based on freedom of speech, or that satire depends on aping reality. But these liberties always come with constraints, and we should not be shy about saying so. That’s how it is today, and has been in the past.

Failure to act against deepfake technology could plunge our society into a state of ‘reality apathy’, where we believe nothing; or chaos, where we are even more likely to believe what is simply false. Let’s learn from our mistakes with social media. This time, let’s act before the horse has bolted.