The sharing of non-consensual intimate images (NCII) has long been a serious issue, disproportionately affecting women, for some time. In recent years it has acquired a new and more alarming dimension with the creation and sharing of “deepfake” NCII – AI-generated imagery which sexualises or degrades identifiable individuals – whether for sexual gratification or simply to victimise, harass and otherwise harm. Media coverage of this phenomenon has grown dramatically,(1) reflecting the scale and speed of the harm.

Online platforms are at the heart of this upsurge. General-purpose AI tools make the creation of deepfake NCII easy (until fairly recently only accessible to experts with expensive software), and users can often circumvent safety guardrails or encounter platforms on which such safeguards are entirely absent. Social media platforms then allow deepfake NCII to circulate at scale – most visibly on X – but smaller services have begun attracting their own regulatory scrutiny(2) for the same reason. The problem is not incidental to the architecture of the modern internet: it is enabled and amplified by it.