Preventive Pathways: How warning messages interrupt and prevent online child sexual abuse at scale
top of page

Preventive Pathways: How warning messages interrupt and prevent online child sexual abuse at scale

Published: 5 June 2026

Preventive Pathways: How warning messages interrupt and prevent online child sexual abuse at scale

Findings and recommendations from a large-scale experiment



Full design, analysis, and results are available in the accompanying research article (arXiv)

CSAM perpetrator research report

Cite this report: Protect Children. (2026). Preventive Pathways: How warning messages interrupt and prevent online child sexual abuse at scale. www.protectchildren.fi/en/post/preventive-pathways-csam-warnings-research-report.



Research shows how warning messages can interrupt and prevent online child sexual abuse at scale


This report presents findings from one of the largest real world studies of warning messages used to interrupt searches for child sexual abuse material (CSAM) online.


The report examines how warning messages shown during CSAM related searches can influence behaviour and redirect individuals towards anonymous support resources. The findings demonstrate that even small changes to warning message design can create meaningful impact at scale.


Why this research is needed


Child sexual abuse material causes devastating and ongoing harm to children, victims, and survivors. Despite growing efforts to combat online child sexual abuse, access to CSAM remains widespread across both mainstream and privacy focused online environments.


There is increasing recognition that prevention efforts must intervene earlier, at the point where individuals search for or attempt to access CSAM online. Warning messages are one increasingly used approach. These messages aim to interrupt harmful behaviour and redirect users towards support before further harm occurs.


However, despite widespread deployment across digital platforms, there has been limited evidence on whether warning messages work, which types of messages are most effective, and how they should be designed.


This report addresses that gap through a large-scale behavioural experiment conducted on the Ahmia.fi dark web search engine, alongside a complementary survey of individuals searching for CSAM online.


Quick findings: 


  • Over a 140-day period in 2025, there were more than 3 million blocked CSAM-related searches on the Ahmia.fi dark web search engine. This is approximately 22,000 per day and represents around 16% of the 20 million total searches. 

  • Each blocked search triggered a warning message linking to an anonymous help resource. 

  • Overall, 12% of warning messages resulted in a click to a help resource, equating to more than 380,000 clicks in total. 

  • We tested multiple warning messages and found that all outperformed a neutral message. 

  • After redesigning and rotating the warning messages, clicks on the link to get help nearly doubled. The average click-through rate increased from around 9% to around 16%. 

  • We surveyed over 6,000 CSAM seekers and found that the impact of warnings extends beyond immediate clicks. Among CSAM seekers who had seen a warning, 17% said they clicked for help or information, and three in four reflected on or changed their behaviour. 


Why the findings matter


The findings show that warning messages can meaningfully interrupt harmful online behaviour and strengthen pathways towards support at scale.


The report highlights that:

  • warning messages are a low cost and scalable prevention tool

  • message design meaningfully influences behaviour

  • warning messages remain effective even in high anonymity environments

  • prevention efforts should focus not only on deploying warnings, but on continuously evaluating and improving them


The findings also underline the importance of investing in broader prevention pathways, including anonymous support services and evidence-based digital interventions.


Whole system recommendations


The report sets out recommendations for governments, regulators, industry, civil society organisations, and the child protection sector.


Key recommendations include:

  • recognising warning messages as a core component of CSAM prevention strategies

  • continuously evaluating warning message effectiveness

  • regularly optimising and rotating messages to reduce habituation

  • simplifying interfaces to maximise engagement with support

  • expanding deployment across digital platforms and environments, including high-privacy and end-to-end encrypted 

  • improving the precision of systems used to trigger warnings


Learn more about Protect Children’s research


Protect Children conducts innovative research on online child sexual abuse and exploitation to strengthen prevention efforts, improve child protection measures, and develop evidence based interventions.





This report was produced as part of a project led by Protect Children and funded by the Home Office.


 
 
bottom of page