Northwestern researchers examine potential harm to democracy
Will the appeal of deepfakes prove irresistible to democratic governments? What questions should governments ask (and who should ask them) when considering a deepfake?
Two Northwestern University professors co-authored a new report that examines several hypothetical scenarios in which democratic governments could consider using deepfakes to advance their foreign policy goals and the potential harm this use could pose to democracy.
He report was published today (March 12) by Center for Strategic and International Studies (CSIS).
The co-authors are VS SubrahmanianWalter P. Murphy Professor of Computer Science at Northwestern's McCormick School of Engineering and a partner in Buffett Institute for Global Affairs; and Daniel W. Linna Jr.full professor and director of law and technology initiatives at the Northwest Pritzker School of Law. They led the report with Daniel Byman, principal investigator of CSIS's Transnational Threats Project.
> Listen: Professors Subrahmanian and Linna discuss the report on the Lawfare podcast
Deepfakes, a digitally altered video, photo or audio recording, are often used maliciously to spread misinformation and create confusion. A well-known example includes a deepfake video that surfaced in March 2022, in which a digitally altered version of Ukrainian President Volodymyr Zelensky tells his soldiers to lay down their weapons.
"As AI has improved, deepfakes have gone from primitive to highly realistic, and will become increasingly difficult to distinguish," the authors write in the report. โThis proliferation of AI provides an unparalleled opportunity for state actors to use deepfakes for national security purposes."
Researchers posit that the lure of deepfakes will eventually become irresistible to democratic governments. โIt won't be long before major democracies, including the United States, begin or at least consider using deepfakes to achieve their ends, if they haven't already." they said.
According to the authors, officials should consider several factors when considering the use of deepfakes:
- the probable effectiveness of deepfake,
- your audience,
- potential damages,
- the legal implications,
- the nature of the objective,
- the target of the deepfake, and
- the traceability of the deepfake to the democratic government of origin.
In general, the authors argue that deepfakes should not be used, as they are likely to reduce the credibility of democratic governments. However, there may be exceptional circumstances where the use of deepfakes warrants serious consideration. In these cases, the authors say, governments should develop a process for approving or rejecting deepfakes that ensures a wide variety of perspectives are put on the table.
