Generative AI tools are getting more powerful by the week, and they're now a routine part of how content is created online. What started as a novelty – funny pictures of politicians, selfies transformed into faux oil paintings – has evolved into darker territory. One of the most sensitive developments is the ability to generate sexual images of real people, so-called 'undressing', without their consent. The practice raises questions that straddle long-standing concerns over image-based abuse, as well as debates surrounding AI governance and the age-old question of freedom of expression.
Research from the Center for Countering Digital Hate found that Grok, the chatbot from Elon Musk's xAI, generated an estimated three million sexualised images in just 100 days, including 23,000 images of children. Cases like these have pushed the issue of AI-based sexual abuse firmly onto policymakers' desks.