A Short Take On Deepfakes

Our laws and legal system have always struggled to keep pace with technology. From cryptocurrency to self-driving cars, legal and regulatory gaps emerge each time technology progresses.

The latest challenge is "deepfakes"—videos created by artificial intelligence and deep-learning technology that either superimpose an individual's face onto another person's body or manipulate existing video and/or audio footage of a person. The result is a fake—but realistic-looking—video that depicts an individual doing or saying things that they have never done.

Since their popularization in 2017, deepfakes have been employed for a variety of purposes. In the political arena, for example, they have been used satirically to provoke dialogue on key issues—like the video produced by a Belgian political party in 2018 of Donald Trump crudely encouraging Belgium to withdraw from the Paris Climate Agreement. Used this way, deepfakes serve a useful social purpose and are protected by Canadian copyright laws, which recognize satire and parody as important and critical tools.1 But deepfakes have also been used in nefarious ways that raise questions about privacy and control in the era of artificial intelligence.

Such concerns were clearly illustrated in November 2017, when a series of deepfakes that superimposed celebrities faces into pornographic videos were released. Importantly, these videos were created and disseminated without the subject's consent. Many of the videos have not been removed and others like them have been created, despite some websites' promises to remove them.

The development of deepfakes has become easier in the past year, with the development of new "user friendly" online deepfake platforms. In this era of technological change, deepfakes present significant risks to both individuals and businesses, and it is presently unclear whether Canadian laws are equipped to provide recourse to deepfake victims.

Personal privacy

The availability of civil causes of action for individual victims of deepfakes has not yet been tested in Canada. Existing law surrounding the nonconsensual disclosure of intimate images (or NCDII—sometimes called "revenge porn") provides a useful starting point. However, despite the similarity of subject matter, the fact that deepfakes often use public photos in fictitious situations presents a challenge for plaintiffs.

Intentional infliction of emotional distress

A claim for intentional infliction of emotional distress is the most promising privacy tort for individual victims.2

A claim for the intentional infliction of emotional distress will be established where there is conduct that a) is flagrant and outrageous, b) was calculated to produce harm, and c) results in a visible and provable injury.3 This test does not require that the defendant be motivated by malice; a claim may succeed where it is clearly foreseeable that posting such a video would result in harm.

The first two elements of this test are established where pornographic deepfake videos are distributed without consent. The sexually explicit nature of the videos...