The Link Between Explicit AI-Generated Images and Offline Crime

Dung Hoang/theispot.com The online revolution enabled frictionless access to virtually every type of content — including adult content. Generative AI has added a whole new, and potentially more dangerous, dimension to the boom in explicit content. In the absence of strong safeguards, AI users can transition from consumers to creators of their own explicit media with shocking ease and speed. National governments around the world as well as the U.N. have issued grave warnings about wide-scale production of AI-generated illegal deepfakes, including images involving minors. Yet Big Tech does not seem to be responding with the same level of seriousness. Earlier this year, Sam Altman, CEO of OpenAI, announced his intention to introduce an “adult mode” within ChatGPT that would allow users to carry on X-rated conversations with the chatbot. The ensuing furor led to the idea being sidelined — for now. Yet Altman’s moment of candor signals that tech leaders are potentially still toying with adult content as part of AI’s value proposition. Our ongoing research suggests that the human consequences of such a decision may be greater and more direct than the “tech bros” expect. Opening the floodgates to AI-generated adult content could, in fact, invite sharp increases in severe sexual criminality. We took advantage of a natural experiment in response to a schism in the scholarly literature. Some psychology researchers theorize that fantasies in online environments have a cathartic effect for potential criminals, keeping their transgressions confined to the digital sphere. But an opposing strain of social learning and stimulation theory has it that online fantasies stoke rather than satiate the desire to commit sexual crimes. We had a unique opportunity to explore this question due to the fact that, unlike many other countries, Japan’s National Police Agency (NPA) publishes detailed monthly crime statistics, including categories of offenses and ages of victims. We accessed NPA statistics for the period January 2022 to October 2024. Smack in the middle of this time frame was the mid-2023 Japanese launch of ChatGPT’s mobile application. We treated that launch as an external shock that materially reduced access frictions in a mobile-first market. (In 2023, 72.9% of internet users in Japan accessed the web via smartphones.) That same period also saw major improvements in image-generation tools, including a July 2023 Stable Diffusion update that made high-quality image creation easier for novice users. So it made sense to use mid-2023 as a technological turning point for before-and-after comparisons. Our analysis found that Japan experienced a significant increase in reported rape cases after mid-2023, with the monthly national average rising from 148 to 265 between the first and second halves of our observation period. Using difference-in-difference techniques designed to create clean comparisons between rape as the “treated” offense and murder as a “control” offense (chosen because both are classified by the NPA as serious violent crimes and follow broadly comparable institutional processes), we found that the increased incidence of rape consistently outpaced that of murder over the same period by approximately 12 cases per month. Before mid-2023, there was no statistically significant difference in the monthly fluctuations of the two types of offense. The same basic pattern held when we compared rape to the other offenses Japan classifies as serious violent crimes that are not necessarily sexual in nature, such as arson and robbery. Still, it could have been a case of correlation rather than causation. Viewed purely in terms of the timeline, the unfortunate spike in rape cases could have been related to all sorts of social factors unobservable in our data. To strengthen our analysis, we obtained data from one of Japan’s largest and most popular online art communities, where users can upload, view, and vote on publicly shared artworks. The community attracts approximately 3.6 million monthly visitors and roughly 2.9 billion monthly pageviews. The platform’s primary language is Japanese, and external web traffic estimates indicate that roughly two-thirds of visits originate from Japan (whose population numbers approximately 130 million). Given this scale and concentration of Japanese users, the platform provides a meaningful window into national-level shifts in online behavior. Importantly, the online community does not prohibit images that are adults-only and/or algorithmically generated, but community members are empowered and encouraged to tag images as such, whether or not they created or uploaded the image. We discovered that periods of peak engagement (likes, bookmarks, and views) with material tagged as AI-generated and adults-only were also those in which the increases in rape relative to murder were particularly high. This statistical relationship manifested only after mid-2023, when generative AI became broadly available in Japan. The totality of this evidence points to a scenario in which AI-generated explicit content directly contributed to a rise in sex-related crime. Rape is one particularly heinous example of such a crime, but there is no reason to expect that this scenario would not also apply to other sexually motivated offenses. The implications of our results became especially alarming when we narrowed the focus to content and criminality involving minors. Engagement with explicit images that were tagged by users as featuring minors was associated with an even larger increase in rape relative to murder. When we zeroed in on rape cases with victims under the age of 20, the estimated amplification was larger still. Although these results do not establish a one-to-one pathway from specific images to specific offenses, they are consistent with a disproportionate increase in offenses involving younger victims. Subsequent studies clarified that the additional crimes mainly consisted of adults offending against minors rather than minor-on-minor crimes. Our Japan-specific findings appear to dovetail with patterns noted elsewhere. For example, observers were shocked and stymied when the CyberTipline, a reporting system for online child sexual exploitation run by the National Center for Missing and Exploited Children in the U.S., announced that in 2023, reports had gone up 12% over the previous year. The figure’s correlation with the timing of the diffusion of AI tools for the general public is highly suggestive of a connection in line with our findings. To be sure, our research merely documents an apparent causal connection. Our findings do not reveal exactly how the dissemination of, and engagement with, AI-generated adult images fueled criminal behavior, nor can we state with confidence that these images were solely responsible for the increase. Certainly, we do not claim to have definitively settled scholarly disputes over whether explicit online content acts as a catalyst for criminal behavior or facilitates an essentially harmless cathartic release. However, the context-specific connection we have identified is serious enough to justify concern about Big Tech’s inclination to integrate explicit content into popular AI models. Against the seeming rationality of Altman’s avowed aim to “ let adults be adults ,” we must juxtapose the probable impact on real-world sexual crime, especially as it pertains to the most vulnerable among us: underage victims. So far, the most aggressive regulatory attention in this area has targeted “revenge porn” deepfakes, as is the case with the TAKE IT DOWN Act signed into law by President Donald Trump in 2025. When it comes to AI-generated explicit content that does not depict identifiable individuals (including content involving minors), current legal frameworks have not caught up to the generative capabilities of AI. As of this writing, for instance, European Union legislators are mired in debate, weighing online privacy c