Privacy, harm, and illusory genitals: A critique of Australia’s deepfake pornography laws

Julian Koplin (Monash University)

Join Zoom meeting:

https://monash.zoom.us/j/86351045263?pwd=1gHMLhmDnXiFJIV0Jl8s6GxhgBgylb.1 

Meeting ID: 863 5104 5263 // Passcode: 184791

Abstract:

The Australian Government recently amended the Criminal Code Act 1995 to explicitly prohibit the distribution of deepfake pornography. It did so by creating a new offence for the transmission of sexual material without the consent of the person depicted. This category encompasses both traditional ‘revenge pornography’ (real sexually explicit content shared without consent) and deepfake or AI-generated pornography (fabricated sexually explicit content.) This legal change was justified on several grounds, each of which is echoed in the broader literature: that AI pornography is harmful, that it involves a privacy violation, that it has harmful social effects, and that it is a gendered practice with disproportionate effects on women. 

In this talk, I tentatively challenge the second, third, and fourth justifications. In particular, I argue that there cannot be a genuine privacy violation when the depicted event never occurred and where an authentic depiction of the individual in question would differ significantly from the fabricated one. AI pornography lacks the verisimilitude required to plausibly claim a privacy violation.

One crucial justification for prohibition remains: being depicted in AI pornography can, and very often does, cause severe emotional, psychological, and reputational damage.  These harms certainly provide grounds for intervention, but I argue they do not justify placing revenge pornography and AI pornography within the same conceptual or legal category, as the Australian law now does. It is plausible to think that the harmfulness of AI pornography is exacerbated when these ersatz depictions are believed, by victims and others, to genuinely reveal private information. I argue that placing revenge pornography and AI pornography in the same legal category blurs the boundaries between them in ways that could plausibly exacerbate the harmfulness of the latter practice. While both forms of pornography are harmful, they require distinct responses. 

This analysis has implications for jurisdictions considering whether to follow Australia’s response to deepfake pornography. It also has implications for how we should conceptualise, discuss, and respond to the problem(s) of deepfake pornography beyond law reform.