Key Insights:

• AI is increasingly being exploited to generate realistic child sexual abuse material, posing new challenges for global law enforcement.
• The IWF has observed a surge in AI-manipulated and wholly AI-generated CSAM, prompting calls for urgent legislative action.
• Enhanced AI technology enables predators to create more realistic CSAM, underlining the need for stronger regulations and public awareness.

Advancements in artificial intelligence are being exploited by paedophiles to create AI-generated videos of child sexual abuse, as reported by the Internet Watch Foundation (IWF). The safety watchdog has raised concerns about a potential increase in the volume of such materials as AI technology becomes more advanced and accessible.

Dan Sexton, Chief Technology Officer at the IWF, expressed worries that AI tools might follow the same trend as AI-made still images, which have surged in volume with improvements in technology and accessibility. He stated, 

“If it follows the same trends, then we will see more videos,” adding that future videos could be of “higher quality and realism.”

Methods of Creation and Current Trends

Most AI-generated child sexual abuse material (CSAM) cases observed by the IWF involve the manipulation of existing CSAM or adult pornography. In these instances, a child’s face is superimposed onto the footage. The IWF noted a few examples of entirely AI-created videos, typically lasting around 20 seconds. 

Analysts from the IWF have seen such videos on dark web forums frequented by paedophiles, identifying nine partial deepfakes where AI models are used to impose a child’s face on existing content.

Although a smaller number of wholly AI-generated videos have been identified, these are currently of more basic quality. Analysts warned that this could represent the lowest quality of fully synthetic videos, implying future versions could be more sophisticated.

Increasing Realism and Prosecution

The IWF has observed a marked increase in the realism of AI-generated CSAM images over the past year. A snapshot study of a single dark web forum revealed 12,000 new AI-generated images posted over a month. The IWF reported that 90% of these images were so realistic they could be prosecuted under UK laws covering real CSAM.

The watchdog operates a hotline for the public to report abuse and has found instances where offenders are selling AI-generated CSAM images online, substituting them for non-AI CSAM. Susie Hargreaves, Chief Executive of the IWF, highlighted the dangers posed by generative AI tools, stating, 

“Without proper controls, generative AI tools provide a playground for online predators to realize their most perverse and sickening fantasies.”

Calls for Legislative Action

In response to these developments, the IWF is advocating for changes in the law to criminalize the creation of guides for generating AI-made CSAM, as well as the production of “fine-tuned” AI models capable of creating such material. Baroness Kidron, a cross-bench peer and child safety campaigner, had tabled an amendment to the proposed data protection and digital information bill that aimed to criminalize the creation and distribution of these models. 

However, this bill was stalled after the general election was called in May by Rishi Sunak.

The urgency of legislative action is underscored by recent reports indicating that AI-generated CSAM is overwhelming US law enforcement’s capacity to identify and rescue real-life victims. The increasing prevalence and realism of such materials pose significant challenges for authorities worldwide.

The IWF’s findings reveal a troubling trend in the use of AI for creating child sexual abuse material. As AI tools become more widespread and easier to use, the production of both manipulated and wholly AI-generated CSAM is likely to increase. This trend not only exacerbates the challenges faced by law enforcement but also highlights the need for robust legal frameworks to address the misuse of AI technologies.

The call for legislative changes underscores the necessity for swift and decisive action to combat the rise of AI-generated CSAM.