Virgin Media’s Deepfake Ireland was never going to be an easy watch. The subject matter was skin-crawling: the growing problem of deepfake and nudification images that are primarily being used as a weapon against women and girls.
Seeing it laid out in such a stark manner was sobering.
The programme initially focuses on presenter Gráinne Seoige’s own experience of having a deepfake image created of her while she was campaigning in the general election in 2024. Two years on, the distress it caused was evident.
But things got worse. With the help of Ceartas, a tech company that started as a free service to tackle so-called revenge porn, she found out the real extent of the problem. A visibly upset Seoige was told that they had uncovered more content, images she hadn’t known about.
READ MORE
There were stories from others who had been similarly affected – model and presenter Maria McGuigan, poet Noel Fleming and an unnamed 15-year-old girl – each one horrifying in its own way.
The message was clear: this isn’t an isolated issue, and it isn’t going away any time soon.
By the end of the year, nudification apps are expected to be banned in the EU. But it has taken far too long to get to this point.
Nudification apps were thrust into the spotlight in January, when images of mainly women and sometimes underage subjects that had been digitally undressed by xAI’s Grok chatbot were published on social media platform X.
If you needed any other reason to remove yourself from the platform, the company’s response – first ignore it, then make access to the image-editing feature available only to paying subscribers – might have pushed you over the edge. Eventually, X cut off Grok’s ability to alter images of real people in places where it breached the law, but only after heavy pressure from authorities. The move came too late for women – and children – who had been affected by the incident, and is not a watertight solution to the problem.
[ Data Protection Commission investigates X over ‘nudification’ of images via GrokOpens in new window ]
While the Grok incident may have highlighted just how easy it is to fake such images, it wasn’t the first to do it, nor was it the worst offender. It was just the most visible. I used an AI image editor to change an innocuous photo of me, fully clothed at a sports event, into one where I was posing in a skimpy bikini. The software refused to do anything more extreme than that, but it is enough. The image editor was a free tool that surfaced in a quick Google search; imagine what you could find if you were willing to spend time to hunt down the most lax software out there. Another few clicks brought me to an AI girlfriend creator and a NSFW (not safe for work) image generator.
This is just scratching the surface of the issue. As the analogy goes, if you see two ants in your home, you don’t have a two-ant problem. Somewhere there is a nest, and until you deal with that, you will continue to have problems.
The difficulty here is that it is almost impossible to root out the nest. While the EU may ban nudification apps and Irish law can take action against those who share nonconsensual explicit images, it cannot stamp out the technology behind it. Apps can be – and are – removed from app stores to make them less visible. But as we already know, things can live on in the quieter corners of the internet, just out of sight but still accessible. As AI skills grow, AI models can and are being modified by users to fit their needs.
Unless there is a concerted global effort to deal with the issue, the apps and those behind them will simply move out of the law’s reach.
That is assuming the ban is passed without being watered down or challenged in some way. In Minnesota, where similar action on such software has been almost unanimously backed by politicians, the proposed law is currently on hold after xAI brought a legal action against it, claiming it imposes an “overbroad, content-based ban on free speech and the tools of visual expression”.
Let that one sink in for a minute. An AI company, owned by a billionaire, claims that a law that bans apps and websites using AI to create nonconsensual sexualised images and videos of people infringed free-speech rights. Even where the political will is there to do it, there are challenges to overcome.
Who asked for this? Why do we need it? The AI companies that have enabled this, intentionally or not, need to take a long, hard look at themselves in the mirror. I once asked an executive in a tech company why, given the obvious potential for abuse, the company was developing tools to generate increasingly realistic images.
The answer? Because the company’s competitors would do it.
The one weapon we have to fight back is education. The idea that the image is fake so it doesn’t have an impact or any legal repercussions is being eradicated by programmes aimed at educating children. Perhaps we should focus on adults, too. Make “I didn’t know” an invalid defence against creating the images.
And while we are at it, erase “we didn’t mean it” from the tech companies’ book of excuses.
