United States of Deepfake 

Abstract

Can the government deepfake a citizen to make it seem as if that citizen endorses a particular viewpoint? Can the government flood our channels of communication with fake bots and messages seemingly from humans but actually from fake accounts in order to influence public opinion, so as to convince the public that a political opponent stole an election, or that a particular military action is justified? Can the government discreetly coordinate with private actors in order to achieve these ends? I argue that for all three scenarios, the answer is “yes” under current constitutional theory.

Perhaps surprisingly, certain constitutional guarantors such as due process and freedom of speech fail to adequately fend off governmental deepfakes and misinformation. Problematically, generative artificial makes it possible to create extraordinarily realistic deepfakes of actual humans, and the technology also exponentially increases the ease by which bots and artificially appearing “humans” can inject themselves into the marketplace of ideas by peddling the counterfeited goods of fabricated speech. Which all begs the question: if generative AI is a weapon, as some scholars assert that it is, what happens when the government deploys that weapon against its own citizens? Should citizens enjoy some type of constitutional right in their identity? If so, what are the contours of such rights? Should the government be prohibited from inserting communications produced from generative AI into our nation’s dialogue? Further, practically speaking, how can such untoward governmental activities be identified and punished?

This Article proposes that for purposes of First Amendment compelled speech jurisprudence, a citizen’s digital name, image and likeness should be considered an extension of a speaker’s person, as well as that speaker’s property. Moreover, in order to guard against the government surreptitiously flooding our communications with fake bots and messages, I propose legislation that provides for whistleblower protection for governmental employees and contractors who report such activity, and further assert that all governmental communications involving generated bots be subject to certain labeling requirements.

Keywords: deepfake, ai, generative ai, artificial intelligence, law