Seeing the Round Corners

HEADS UP, the new day for Seeing the Round Corners “GOING LIVE” is Tuesday each week.

October 29, 2019


Readers most likely realize by now, deep fakes and deep fake videos are a really complicated subject matter. It is a subject matter that the consideration of consequences have only begun. As seen over the decades of technology advancement has given the world innovations at an unstoppable pace, too many to count. Just as with deep fakes, many innovative inventions came with the potential to harm and no capacity to debunk or way of limiting potential harm. 

There is some AI (Artificial Intelligence) developed and trained by GIF hosting company Gyfcat to spot fraudulent deep fake videos. Project Maru “can spot deep-fake videos because in many frames, the faces aren’t perfectly rendered, and Project Angora masks the face of a possible deep fake, and searches the internet to see if the body and background footage exist elsewhere.” So there is some availability for rapid detection of deep fakes. 

The small number of technology solutions of deep fakes and deep fake videos is fertile ground for grant-making agencies such as the National Science Foundation (NSF) and the Defense Research Projects Agency (DARPA) to “begin steering funds towards scalable detection systems that can be commercialized or even provided freely.” The dearth of solutions may also “generate new market forces encouraging companies to invest in such capabilities on their own or in collaboration with each other and academics.”

Legal solutions to deep fakes and deep fake videos are just as complicated as the subject itself. Chesney and Citron state:  No current criminal law or civil liability regime bans the creation or distribution of deep fakes. A threshold question is whether such a law would be normatively appealing and, if so, constitutionally permissible. A flat ban at this stage of the game is doubtful according to Chesney and Citron because “deep fakes exact significant harm in certain contexts but not all. A prohibition of deep fakes would prohibit routine modifications that improve the clarity of digital content. It would chill experimentation in a diverse array of fields from history and science to art and education.” (Bobby Chesney and Danielle Citron are authors of Deep Fakes:  A Looming Challenge for Privacy, Democracy and National Security.)

Perhaps the scariest aspect of a general prohibition of deep fakes will cast a “significant shadow, chilling expression crucial to self-governance and democratic culture. Both unpopular or dissenting views could be censored by government officials. The free speech folks would most likely push to forego an outright ban of deep fakes than to run the risk of abuse.

Chesney and Citron state:  Deep fakes implicate freedom of expression, even though they involve intentionally false statements,” and refer to The First Amendment Overbreadth Doctrine (describing the presumption that courts have against statutes that curtail a broad array of expressive activity).

And the high court (U. S. Supreme Court) weighs in:  “. . . false speech enjoys constitutional protection insofar as its prohibition would chill truthful speech.” In a later case, Alvarez, the U. S. Supreme Court clarified:  “. . . barring the application of an exception to the First Amendment’s protections, false statements can be regulated only insofar as defendants intend to cause ‘legally cognizable harm’ and a direct causal link existed between the ‘restriction imposed and the injury to be prevented.’”

Also in Alvarez the Court acknowledged “certain categories of speech are not covered by the First Amendment due to their propensity to bring about serious harms and their slight contribution to free speech values.” These categories are where some deep fakes fall. Chesney and Citron categorize unprotected lies to include the defamation of private persons, fraud and impersonation of government officials. The First Amendment affords no protection to “speech integral to criminal conduct and the imminent-and-likely incitement of violence.”

Next week, more on civil liability.

The reader's comments or questions are always welcome. E-mail me at