Digital forensics in the UK is constantly challenging. The volume, complexity and technological sophistication of child sexual abuse material (CSAM) is set to surpass the capacity of existing systems, legislation and guidance to respond proportionately. Throughout a series of workshops involving digital forensic practitioners, investigators, offender managers and representatives from the wider criminal justice system, one theme was reiterated with striking clarity: “we cannot continue as we are”.
The absence of national frameworks, uneven understanding of accessibility and offender risk, inconsistent thresholds, and the accelerating emergence of AI-generated CSAM have created a system wholly reliant on professional discretion, regional interpretation and an under-funded workforce. These issues are now compromising not only consistency and fairness, but the ability of the system to safeguard children and deliver justice at pace.
This whitepaper reflects a growing consensus that the UK requires a coordinated, national recalibration of digital forensic responsibilities, standards and expectations, supported by modernised legislation, refreshed Home Office guidance and a sustainable operational model.
1. Introduction: A System Under Unsustainable Pressure
The policing response to CSAM has evolved rapidly over the past decade, but its foundational structures have not kept pace. Digital forensic units (DFUs), originally designed to extract and interpret information from devices, are now routinely tasked with functions far beyond their intended remit: assessing sexual acts, estimating ages of subjects, undertaking categorisation, and producing risk-relevant material for offender managers. These are roles traditionally associated with safeguarding specialists, psychologists, paediatricians, or investigators, but, in practice, are increasingly absorbed by digital forensic practitioners by default rather than design.
The workshop discussions made it abundantly clear that practitioners are working against an infrastructure that has not adapted to the realities of modern offending. The result is a system fragmented in practice, slow in output and increasingly vulnerable to legal challenge.
2. Streamlined Approaches to Low-Risk Offenders: Innovation Amid Inconsistency
Several forces have attempted to mitigate increasing workloads through innovative approaches. One force, for example, has developed a local agreement with its Crown Prosecution Service (CPS) area that allows triage reports to be accepted as evidential submissions for first-time offenders presenting low safeguarding risk. This has significantly reduced device backlogs and accelerated the criminal justice process, while also enabling police online investigation teams to progress cases before devices are passed into DFU queues.
Yet this promising approach also exposes a deeper problem. There is no national definition of a “low-risk offender,” no shared criteria for its assessment, and no agreed risk-management model governing when triage evidence is appropriate. Forces therefore operate in isolation, compelled to develop their own definitions and methodologies, which inevitably produces regional disparity.
A second concern raised within the workshop relates to evidential scope. If a suspect is charged on the basis of a triage-derived report and further material is subsequently identified during full examination, practitioners questioned whether the system enables the appropriate addition of further charges, and whether triage evidence might later constrain investigative or prosecutorial flexibility. Without clear national guidance, practitioners are left to navigate these questions case by case.
This area alone illustrates how inconsistency has become structurally embedded within the system.
3. Categorisation Systems and the Increasing Impossibility of Manual Review
The long-standing A/B/C categorisation scale continues to function as a core evidential tool, yet a number of participants challenged whether it remains appropriate in an era defined by extreme volume and synthetic imagery. The increasing presence of AI-generated CSAM in particular raises questions about both the feasibility of manual categorisation and the clarity of the underlying legislative intent.
Several practitioners emphasised that manual categorisation on a national scale has become unsustainable. The system is simply not designed to cope with the volume of files, the speed of technological change or the complexity of distinguishing synthetic from real victims. Without reform, the backlog will continue to grow, and the risk is that the highest-harm cases become lost within a queueing system originally designed for an entirely different operating environment.
The workshop also raised the prospect of utilising magistrates’ courts more routinely for possession-only cases for first-time guilty pleas. Participants were keen to understand if this would alleviate Crown Court pressures, deliver resolutions more quickly and reduce overall CPS expenditure. The wider justice implications of this approach merit further exploration, but it is clear that timeliness and proportionality are now critical factors requiring system-level attention.
4. Accessibility: A Concept Lacking National Understanding or Consistency
A particularly challenging theme arising from the second workshop concerned file accessibility and the widespread misunderstanding that surrounds it. Practitioners repeatedly reported that CPS and the judiciary remain several years behind the digital reality, with many unfamiliar with how accessibility should be understood, evidenced or contextualised.
Despite this lack of clarity, the workshop reached a firm conclusion: accessibility must be reported. Failure to do so risks significant evidential challenge and, in some cases, the dismissal of material due to ambiguity or misinterpretation. Yet many forces have moved away from declaring accessibility because they fear challenge, lack clear parameters or are conscious that current tools offer only partial insights. This reticence, however understandable, introduces a significant operational and legal vulnerability.
Offender management teams echoed similar concerns. They highlighted their limited ability to interpret MG22 reports and their need for more nuanced contextual information, including IT literacy, technical knowledge and behavioural indicators. Without sufficient funding or awareness, post-conviction risk management becomes inconsistent, reactive and reliant on incomplete data.
5. Sentencing Volumes: Thresholds, Ambiguity and a System Out of Alignment
The discussion on sentencing volumes exposed fundamental disagreements about what the criminal justice system is attempting to achieve. Thresholds established within CPS guidance were never intended to function as rigid evidential or sentencing triggers; however, they have become de facto targets. This shift has generated significant inconsistency across regions, with forces and prosecutors applying different interpretations and thresholds to cases involving similar material.
Workshop participants questioned the logic of a system that places such emphasis on volume when harm severity, victim age and contextual factors often carry far greater relevance. Analogies were drawn to firearms legislation, in which one weapon is sufficient for charge, raising the question of why CSAM offences retain volume-based thresholds at all.
Underlying this discussion was a more fundamental issue: should digital forensic practitioners continue to be the professionals responsible for categorising indecent images of children? Outside National Accredited Grader training, they are not fully equipped for these determinations, nor should they be expected to absorb the psychological burden associated with repeated exposure to the most traumatic material. Digital Forensic Analysts are currently victims of vicarious trauma. The current arrangement has emerged through operational necessity rather than strategic design, and the consequences for workforce wellbeing are increasingly evident.
6. AI-Generated CSAM: The Current Crisis in Waiting
The fourth workshop highlighted the profound uncertainty surrounding AI-generated imagery. Participants agreed that current legislation does not adequately address synthetic CSAM and that the speed of technological development has far outpaced policymaking. Determining whether an AI image constitutes an offence is increasingly subjective, particularly when images are photorealistic or when no real child has been harmed in their creation. The question of whether the offender recognises the synthetic nature of the material adds further complexity.
A central concern is the widening accessibility of generative AI tools. These platforms no longer require technical proficiency, access to the dark web, or specialist knowledge. They are openly available, often free, and increasingly embedded in mainstream applications. This democratisation of capability means that image creation, whether of a fully synthetic child or manipulated imagery depicting a real child, is now achievable in seconds, challenging long-held assumptions about offender profiles and pathways into offending.
Workshop discussions highlighted the critical distinction between:
1. Fully synthetic AI-generated images, depicting children who do not exist; and
2. AI-enhanced or AI-manipulated depictions of real children, where identifiable minors are transformed, sexualised, or exploited through artificial means.
Both categories present substantial legal and ethical uncertainty. In the second scenario, a real child’s identity, likeness, or personal imagery can be appropriated without their knowledge or involvement. The harm here is tangible – images may circulate indefinitely, and victims face reputational, psychological and safeguarding consequences. This raises urgent questions about whether the presence of a “real child” is required for an offence, or whether the exploitation of a child’s identity alone carries sufficient weight to warrant criminalisation.
The group discussed whether the origin of the imagery (real versus synthetic), ought to form part of a defence. Consensus was clear that this is a deeply problematic area. If an individual knowingly creates, possesses or distributes sexualised images that depict child-like bodies, features, or personas, the intention, sexual gratification, and broader risk profile remain significant, regardless of whether the child is real. The analogy was made to imitation firearms: an armed robbery committed with a realistic fake weapon is still treated as an armed robbery. Harm and culpability derive from behaviour, intent and societal risk, not merely the physical authenticity of the weapon or, in this case, the child.
Participants also emphasised that young people themselves are increasingly generating sexualised AI content, often naively, using their own images or those of peers. This introduces a complex safeguarding dimension: behaviour that may not be predatory or malicious can nonetheless result in criminality, victimisation, and long-term consequences. Schools, parents, social care and policing are already struggling to respond to a generation that has access to powerful AI tools without fully understanding the legal ramifications or personal risks.
There was strong caution expressed against any reliance on AI systems to make determinations within the forensic process. While AI may support the identification of potentially relevant material, it cannot yet be trusted to replace human oversight. Over-reliance on automated classification risks misidentification, bias and significant miscarriages of justice, and the forensic community remains understandably resistant to commercial claims that AI can fully “replace” human examination.
7. Cross-Cutting Themes: A System in Need of Coherence
Across all discussions, several unifying concerns became apparent. The system lacks a coherent feedback loop between DFUs, FSPs, investigators, CPS and offender management teams, leaving each group to operate with a burden and a lack of strong mutual understanding. Officers in charge (OICs) of investigations may feel disconnected from the pace of technological change, which leaves them reliant on DFUs, while DFUs themselves lack the contextual offender information needed to make proportionate decisions. The entire system is therefore characterised by a pattern of dependency and fragmentation.
Most significantly, there is no shared national framework guiding proportionality, risk assessment or the allocation of forensic effort. As a result, regional practices diverge widely, contributing to the “postcode lottery” repeatedly noted by participants.
8. A Call for National Reform
The workshop concluded with a clear demand for change. Participants agreed that the UK now requires a nationally mandated steering group to drive the development of consistent, modernised approaches to digital forensic practice in CSAM investigations. It is imperative that steering group participants are those with practical knowledge of operating within a DFU/FSP, rather than those with theoretical. Such a group must draw together the Home Office, NPCC leadership, DFUs, FSPs, investigators, CPS, judiciary, offender management teams and technical experts. Its remit should include the creation of a national framework for categorisation, accessibility assessment, risk-based triage pathways and the appropriate role of AI within the forensic process.
Finally, and perhaps most urgently, the wellbeing and sustainability of the workforce must be addressed. The increasing volume and graphic nature of material, combined with the expectation that practitioners undertake categorisation duties outside their professional expertise, is both untenable and unsafe. The creation of specialist categorisation teams, supported by appropriate clinical and organisational structures, must now be seriously considered.
9. Conclusion
The digital forensic response to CSAM offending is no longer merely under strain; it is operating beyond its intended design. Without decisive national action, the inconsistencies, backlogs and risks identified in these workshops will continue to grow, and with them the danger of failing those we are entrusted to protect.
The message from practitioners is unequivocal. We need coherent national guidance, a structured risk-based approach, modernised legislation and a sustainable workforce model. Above all, we need a unified strategy that reflects the realities of contemporary offending and equips us to respond effectively to those who pose the greatest risk to children.
It is now time for national leadership to match the urgency experienced every day by those working within digital forensics.
