Defamation in the Time of Deepfakes

Abstract

Deepfake technology, powered by artificial intelligence, has enabled the quick and easy creation of hyperrealistic videos that superimpose one person’s face onto another’s body. While the technology has benign applications, it has also been overwhelmingly used to create nonconsensual pornography. Deepfake pornography is a severe sexual offense that has targeted hundreds of thousands of women. This Note, the first comprehensive analysis of deepfake pornography under defamation law, sketches a framework for advocates and judges to apply defamation to cases of deepfake pornography.

This Note argues that deepfakes—in achieving photorealism and simulating someone’s true body and private life—qualify as defamatory false statements of fact. As this Note shows, when alleged defamatory statements strive for (and achieve) hyperrealism, and they purport to reveal a truth about someone’s private sex life, they qualify as false statements of fact. Cursory indications that a deepfake is “fake” or even viewers’ knowledge that it is “synthetic” refer solely to the manner of creation, not its signified meaning. The photovisual realism of deepfakes collapses the distinction between form and meaning or signified and signifier. As signifiers whose forms perfectly resemble their signified, deepfakes leave no room for the person depicted to disavow their message or for the statements to transform into a parody or commentary protected by the First Amendment. Thus, the knowledge that a deepfake is fake does little to undermine the reputational harm and, consequently, the defamation claim. Finally, this Note addresses defamation law’s peculiar and controversial “actual malice” scienter requirement. As actual malice relates to knowledge or reckless disregard for the falsity of the statement and not a defamatory intent, it applies to creator-distributors who use synthetic processes to make deepfakes, albeit often claiming a benign or parodic purpose.