Non-consensual intimate images have been weaponised against women for decades. Now AI has made their production trivially easy and their proliferation almost impossible to contain. The legal system is struggling to keep pace.
In early 2024, AI-generated intimate images of the pop star Taylor Swift circulated on social media platforms, reaching tens of millions of users within hours before platforms scrambled to remove them. The incident prompted a surge of political attention, legislative proposals, and public debate. It was, in some respects, a turning point; a moment when a harm that had been primarily visited upon ordinary women, often without any public attention at all, became impossible to ignore.
The production and distribution of non-consensual intimate images, NCII, is not a new phenomenon. What is called ‘revenge porn’ has existed since the early days of digital photography. The creation of manipulated, ‘deepfake’ intimate images began as soon as sufficiently powerful AI image-generation tools became available, around 2017. What changed in 2023 and 2024 was accessibility: the emergence of consumer-grade AI tools meant that creating a photorealistic intimate image of any person from a handful of photographs became something that could be done in minutes, for free, by anyone with a smartphone.
The Scale of the Problem
The statistics are stark. A 2023 report by the firm Home Security Heroes found that the number of deepfake videos online had grown by over 550% since 2019, and that 98% of deepfake videos online were non-consensual pornographic content, with 99% of subjects being women. A separate analysis by researchers at University College London found that AI-generated NCII had become significantly more difficult to detect as fakes, with generative models producing images that even trained forensic analysts struggled to identify as synthetic.
The victims of NCII, whether real images distributed without consent or AI-fabricated images, report consistent and severe harms: profound distress, anxiety, and depression; damage to personal and professional relationships; loss of employment; and in some cases, long-term psychological trauma. There is a documented correlation between image-based sexual abuse and suicidal ideation. The harms are real, serious, and disproportionately visited upon women, particularly young women, women of colour, and women in the public eye.
Where the Law Stands
The legal landscape is, to use a charitable term, uneven. England and Wales have moved faster than most comparable jurisdictions. The Online Safety Act 2023 criminalised the sharing of non-consensual intimate images without requiring proof of intent to cause harm, a significant improvement over previous legislation. The Criminal Justice Bill, as it progressed through Parliament in 2024, went further, proposing to criminalise the creation of AI-generated NCII even without distribution. This would be among the most progressive positions in the world on this issue.
In the United States, there is no federal law specifically criminalising NCII or deepfake pornography as of the time of writing. A patchwork of state laws exists, with significant variation in definitions, penalties, and the requirement to prove intent. The DEFIANCE Act, introduced in the Senate in 2024, would create a federal civil cause of action for NCII, allowing victims to sue those who create or distribute images; it has broad bipartisan support but has not yet been enacted into law.
Australia enacted legislation criminalising NCII in 2023. The European Union is addressing the issue through a combination of the Digital Services Act, which places obligations on large platforms to address illegal content, and specific criminal law provisions proposed under a directive on combating violence against women. Canada and New Zealand have laws that address aspects of the problem, but comprehensive deepfake-specific legislation remains limited globally.
98% of deepfake videos online are non-consensual pornographic content. 99% of subjects are women. Consumer-grade tools now make creating such images trivially easy. Legislation has barely begun to respond.
The Platform Problem
Law criminalising the creation and distribution of NCII is necessary but not sufficient. The platforms on which images spread are critical actors, and their response has been inadequate. Major social media companies have policies prohibiting NCII, but enforcement is reactive, inconsistent, and dependent on victims finding and reporting content themselves; a process that is often retraumatising and rarely comprehensive.
The ‘whack-a-mole’ problem is real: images removed from one platform reappear on others, often within hours. Dedicated websites hosting NCII operate in jurisdictions where enforcement is difficult. End-to-end encrypted messaging apps present further challenges for detection. The technical solutions that exist (image-matching technology that can identify known NCII and prevent its re-upload, along lines similar to PhotoDNA, which is used for child sexual abuse material) have not been deployed at sufficient scale for adult NCII.
What Justice Looks Like for Victims
Legal reform, while necessary, is of limited comfort to someone whose intimate images are already in circulation. Victims and advocates consistently identify three priorities beyond criminalisation: rapid and mandatory takedown processes from platforms; proactive detection and removal rather than reactive complaint-based systems; and access to support services (including legal advice, mental health support, and technical assistance with removal) without charge.
Organisations like the Revenge Porn Helpline in the UK, the Cyber Civil Rights Initiative in the US, and StopNCII, an image-matching tool that allows victims to protect their images across participating platforms, provide crucial support. But they operate on limited resources in the face of an industrialised harm.
The deeper question is one of culture as much as law. The creation and consumption of NCII is not a technological inevitability: it is a choice, enabled by technology and shaped by attitudes toward women’s bodies, consent, and sexuality that predate the internet by centuries. Legislation can and should make it harder. But changing those attitudes is the longer work, and it starts with insisting that this is everyone’s problem, not merely the problem of the people it happens to.