Deepfake Threat on the Horizon
Michael Chertoff, Former US Homeland Security Secretary; Co-Chair, Transatlantic Commission on Election Integrity;
Eileen Donahoe, Executive Director, Global Digital Policy Incubator, Stanford University; Member, Transatlantic Commission on Election Integrity
As we look ahead to future elections, there will be a far more dangerous tool in the election interference toolkit: deepfakes, which are Artificial Intelligencebased human-image synthesis techniques, that combine and superimpose existing images and video onto source video.
Deepfake technology will enable malign actors anywhere to create video of virtually anyone, doing and saying whatever they want them to. Deepfakes are becoming less prohibitively costly to produce just as they become more convincing. This technology will soon be available not only to malign states, but to malign individual actors.
Imagine what could happen to public trust and civic discourse around elections as this technology spreads. Put bluntly, deepfakes could transform not just election interference, but politics and geopolitics as we know it.
So what can be done to prepare ourselves for the next wave of election interference via deepfakes?
First off, Artificial Intelligence-based detection of deepfakes must be turned against malign actors, so that deepfakes can be quickly identified and stopped before they spread. Artificial Intelligence can now be utilised to sniff-out imperfections in manipulated video invisible to the human eye, through watermarking algorithms and metadata built into authentic video. Development of this detection technology must be job number one.
Private sector platforms should embrace this detection challenge as a shared public interest priority. They should turn their R&D fire onto this urgent threat, before it shows up and spirals out of control on their own platforms. The key here will be to focus on detecting manipulation of source video (not evaluation of political content).
But perhaps most importantly, civic education about the threat of deepfakes must be seen as an essential element of democratic defence against this new generation of disinformation. Governments, civil society and private industry should team up in a massive public education campaign to inoculate the public — before deepfakes spread virally, dramatically impact public opinion, or change an election outcome.
In this vein, we believe strongly that deepfake technology itself, must be utilised to educate the public and showcase the power of this technology. This is why the Transatlantic Commission on Election Integrity is enlisting the help of technologists in building a Deepfakes Civic Education Platform. Our goal is to help citizens become discerning consumers of video and audio material, especially around elections.
The bottom line is that without greater public awareness of the danger, deepfake technology has the potential to cause electoral chaos and geopolitical instability. Democratic governments need to get ahead of this threat by engaging the public in the defence of democracy. Building citizen resilience to deepfake disinformation must become a shared public interest priority.
