The amount of child sexual abuse material (CSAM) generated using artificial intelligence is growing exponentially. We wanted to investigate the topic further to gain a better understanding of what needs to be done from a prevention perspective.
In this report, we examined discussions taking place on dark web forums that focus on child sexual abuse. The analysis of these discussions aimed to determine what participants think about AI-generated CSAM and the AI technologies used to create it.
Key findings of the study:
The general opinion of participants in dark web forums regarding AI-generated CSAM
AI-generated CSAM is created in large quantities, easily, and willingly. There are both people creating and requesting this material. Many comment that the material is better than any real material they have seen before and the best they have ever seen.
Community, praise, and enthusiasm
There appears to be a strong sense of community on dark web forums. It is easy for people creating material to receive feedback on it, most of which is positive. The material is analyzed in detail, and any issues related to its creation are resolved collaboratively. The community seems to support and encourage people creating material to produce more, and the community also maintains motivation and enthusiasm to share expertise.
Styles and Preferences
It is shared that the AI models are trained using everyday images sourced from, for example, social media, as well as real, non-AI-generated CSAM. Real and AI-generated material are intermingled, and users do not always know which is which. AI-generated material typically aims for a result that is as photorealistic and lifelike as possible.
Ethics and Justification
Forum participants view AI-generated material as more ethically sound than real material and believe it protects children—even AI-generated material based on real CSAM is not perceived as harmful to children. Participants consider AI-generated CSAM to be a safer alternative than real material, and they do not believe the law can keep pace with technological developments.
The Future of AI
There is strong confidence in the future development and improvement of AI. People are repeatedly and positively surprised by the current level and capabilities of AI. In the forums, it is believed that what is not yet possible will become possible in the very near future.
Skills in Using AI Technology
Skills regarding the use of AI technologies vary. Some participants are beginners and ask many questions related to the basic use of AI technology. Others are clearly more skilled at using AI and engage in significantly more detailed discussions. The discussions reveal that an individual does not need to be technically skilled, as the community is.
The quality of AI-generated CSAM
There are many discussions about the quality of the material, with participants sharing experiences to improve it. When discussing the quality of the material, topics range from technological solutions to the use of various AI models, as well as their training and fine-tuning. These discussions seem to encourage participants to produce material of ever-higher quality.
With AI, anyone can become a subject
Advances in large image models have enabled improvements in the performance of AI image generators, both in terms of usability and the quality of the final output. These advancements have led to a situation where an increasingly small amount of source material is now sufficient. For example, images of a known victim of sexual abuse or a child in the public eye are enough to generate convincing CSAM of that person.
What’s next?
The report will be published on this page in October 2026.
Several recommendations for different target groups have been created based on the report. These recommendations will be published as an appendix section to the report.
If you are interested in collaboration related to the content of the report, please contact Nettivihje: nettivihje.viestinta@pelastakaalapset.fi
