The EU Parliament is pushing to make AI-generated child sexual abuse material, or AI-CSAM, explicitly illegal under a new directive. The move comes as the Internet Watch Foundation (IWF) warns that synthetic abuse content is growing quickly and becoming easier to create.
The issue is not only that harmful material can now be generated with less technical skill. The IWF also warns that AI-created CSAM can complicate efforts to identify real children who are being abused, adding pressure on lawmakers to close gaps in the law.
What the IWF says is changing
According to the IWF, the first confirmed case of AI-generated CSAM appeared in 2023. In 2024, reports rose by 380%, reaching 245 incidents involving more than 7,600 images and videos.
The organization says the most severe category of abuse, Category A under UK law, accounts for nearly 40% of AI-generated CSAM. That is almost double the proportion seen in traditional cases, according to the source article.
The material also overwhelmingly depicts girls. About 98% of synthetic material depicts girls, compared with 97% seen across all forms of CSAM.
The IWF says offenders are using tools such as text-to-image generators and “nudify” apps. The most advanced AI systems can also produce hyper-realistic short videos, making the content more convincing and more difficult to distinguish from real imagery.
“What we're seeing now is highly realistic abuse imagery being generated with minimal technical skill. This technology is being exploited to cause real harm to children,” said Dan Sexton, the IWF's Chief Technology Officer.
The report also highlights a further concern: in the most disturbing cases, AI models are being trained on real abuse images. That means synthetic material can still be connected to real-world harm, even when the final image or video is generated by software.
What the EU Parliament wants to ban
EU law currently lacks explicit rules on synthetic abuse material. The EU Parliament wants the proposed Child Sexual Abuse Directive (CSAD) to close that gap by fully criminalizing AI-CSAM.
Lawmakers want the ban to include possession for “personal use” and reject exceptions. They are also seeking clearer definitions, better detection tools, and stronger cross-border cooperation for police and child protection agencies.
The Parliament’s position treats AI-generated CSAM as a category that requires direct legal attention, rather than leaving it to be handled indirectly through older rules. That matters because synthetic material can be created, shared, stored, and modified in ways that existing frameworks may not clearly address.
The proposed approach also recognizes that enforcement is not only about punishment after the fact. Detection tools, shared definitions, and cross-border cooperation all affect whether authorities can respond when material moves between platforms, devices, and jurisdictions.
The dispute over “personal use”
The EU Council’s current position has drawn heavy criticism. Its draft version of the CSAD would allow people to possess AI-generated abuse images for “personal use,” a carveout the IWF calls a “deeply concerning loophole.”
The IWF and its partners are pushing to close that gap. Their argument is simple: there is no harmless version of abuse material. Even when a file is synthetic, it can normalize abuse, be used to target children, or make investigations harder.
The “personal use” question is central because possession rules determine whether someone can legally keep this material without distributing it. The Parliament’s stricter stance would remove that distinction for AI-generated CSAM.
The source article also notes another challenge. AI-generated CSAM can make it harder to identify real cases of abuse. Investigators may have to separate synthetic files from real images and videos while still treating every potential victim lead seriously.
Beyond images and videos
The IWF is not only calling for a full ban on AI-generated CSAM itself. The organization also wants an EU-wide prohibition on guides, instructions, and models used to create CSAM.
That demand reflects the way generative AI systems can spread harm before any final image or video is shared. Instructions, trained models, and creation workflows can become part of the abuse ecosystem if they help people generate illegal material.
The organization is also calling for better support for survivors. That point is important because the harm does not stop at the technical question of whether content is real, synthetic, or modified. The existence and circulation of abuse material can continue to affect victims and survivors.
The new directive is still under negotiation. For now, the central disagreement is clear: the EU Parliament wants a strict ban on AI-CSAM, while the Council’s current draft contains a “personal use” exception that child protection advocates want removed.
As AI tools become more capable, the legal debate is moving from whether synthetic abuse material is possible to how directly it should be criminalized. The Parliament’s position is that the answer should be explicit, comprehensive, and without exceptions for private possession.