29 April 2025
Translational Fellow and author: Dr Alice Siberry, Creased Puddle.
- Whilst this study cannot establish definite links between neurodivergence, extremism and algorithms, practitioners identify neurodivergent people as vulnerable to self-radicalisation facilitated by online algorithms.
- A lack of comprehensive training may lead practitioners to misattribute extremist behaviours to neurodivergence, leading to lost opportunities to critically assess vulnerability and risk.
- This study recommends specialist neurodiversity training in Counter Terrorism, supported by a self-directed toolkit, to improve understanding across policing, education, and healthcare agencies.
Summary
This study explored the intersection between neurodiversity, extremism and online algorithms, identifying gaps in practitioners’ understanding.
Practitioners agreed that neurodivergent people, particularly young autistic males, may be vulnerable to self-radicalisation facilitated by algorithms. However, most understanding about neurodiversity was generalised, conflating neurodivergent traits, such as “fixation” (or ‘specialist interest’, an urge to learn everything about that interest) or “social isolation” and extremist behaviours.
Oversimplified narratives risk criminalising neurodivergent people without addressing broader socio-technological factors, such as increased use of social media and algorithmic learning. Mismanagement of risk and lack of engagement from external services, such as schools and social workers, may further exacerbate the contextual vulnerability faced by young neurodivergent people.
To mitigate these risks, this study calls for neuroinclusive, role-specific training and collaborative, solution-focused strategies between Counter Terrorism (CT) practitioners and external services.
Background
Introduction
Neurodivergent traits are often reported as risk factors for radicalisation, a perspective that overlooks broader socio-technological influences like online algorithms. This summary highlights gaps in understanding neurodiversity within CT, regarding how algorithms can influence involvement in extremism and the need for more solution-focused, neuroinclusive approaches to CT strategies and interventions.
Neurodiversity and Counter Terrorism
Mental health vulnerabilities, including neurodivergence, contribute to a significant number of Channel (the multi-agency service supporting vulnerable people at risk of extremism) and Prevent referrals (the country’s safeguarding system for vulnerable people’s involvement in extremist or terrorist activity). Existing studies link autism-specific traits like ‘restricted interests’ and ‘difficulties with social interaction’ to extremist behaviour, whilst neglecting the broader situational and technological contexts that can exacerbate risk. Online algorithms pose a new threat, potentially targeting neurodivergent people through personalised content recommendations.
Lessons learned
Similar to the disproportionate profiling of Muslim communities under the Prevent strategy reported over a decade ago, there is a risk that technological advancements may drive the referrals of neurodivergent people, who may not pose an actual threat. These referrals could reflect societal tendencies to medicalise and pathologise extremism, echoing discriminatory patterns observed in Prevent’s treatment of Muslims under previous strategies. Despite the absence of concrete evidence linking neurodivergence to extremism, the characteristics and traits of autism specifically appear to continue to be linked in a similar way.
More critical works argue that neurodivergent traits should not be conflated with deterministic radicalisation pathways, as has been documented historically in Muslim communities. Addressing these challenges requires a nuanced approach to risk assessment, accounting for the role of online radicalisation. CT frameworks must integrate broader socio-technological and intersectional perspectives to better address the unique vulnerabilities faced by neurodivergent communities.
What we did
The aim of this research was to explore whether police practitioners recognised neurodivergent people as being at risk of exploitation by algorithms.
It also assessed whether practitioners are equipped to support neurodivergent people and understand their involvement in extremist activities. The research sought to explore the broader experience of neurodivergence in extremist contexts, beyond just autistic suspects. The ultimate goal was to propose solution-focused approaches using neuroaffirming language to challenge potentially harmful narratives about a perceived vulnerable group.
Semi-structured interviews were conducted with 10 participants from a regional CT policing unit. Due to the sensitive nature of this area of policing, the roles of each participant have been anonymised. However, there was at least one participant from each of the specialist roles within that particular CT unit, including (but not limited to) Prevent, Nominal Manager, Intelligence and Investigations. Data from interviews were recorded, transcribed and analysed using thematic analysis.
Key findings
Introduction
Practitioners predominantly discussed autistic people when asked about neurodiversity, often using medicalised language. The focus was largely on young autistic males, with minimal references to adults, females or other types of neurodivergence. This reflected interviewed practitioners’ caseloads rather than the broader CT portfolio. Practitioners almost exclusively linked neurodivergent people with right-wing extremism.
The impact of surface-level understanding
Participants agreed that neurodivergent people are vulnerable to exploitation via online algorithms, describing a self-radicalisation process beginning with “innocent” content like memes or gaming, progressing to darker material by “chance”. This shift, involving activities like sharing graphic content in encrypted group chats, was seen in both neurodivergent and neurotypical people.
However, practitioners suggested that autistic traits specifically, such as “fixation” and social isolation, increase vulnerability. One stated, “algorithms play… the most important part. I think when they [autistic people] go on the rabbit hole, they maybe find associations with people online and they draw to become more like those people”.
Despite these observations, understandings of autism appeared surface-level, rarely considering situational risk factors. Practitioners also highlighted challenges discerning whether neurodivergent people intend to act, with some committing online offences without grasping the consequences. As one practitioner noted young people comment: “‘I just said it’ and ‘I didn’t mean it’ and it’s kind of like no, you said it, you’ve posted some horrific things… it’s not all fun and games”. Interviews reflected the sentiment that: “there’s no grievance, there’s no ideology, it’s just neurodiversity”.
Inconsistency in training and partnership working
Minimal neurodiversity training shaped practitioners’ answers, with only five practitioners having had any training. Practitioners shared: “there’s very little [neurodiversity] training that is made available to anyone”. Without training, a connection between the reliance on stereotypical traits to explain actions and motives was established. Better trained practitioners showed more confidence in adjusting practices.
Many practitioners acknowledged that criminalisation of neurodivergent behaviours often stemmed from mismanaged risk and vulnerability, which was compounded by limited engagement with external services like education and social services. These agencies were described as ‘risk averse’. Practitioners expressed frustration over neurodivergent people falling into service voids, with fluctuating risks due to inadequate support and training: “it would be nice to see moving forward in terms of managing the risk that these people [partners] are more focused on actually providing services to those young individuals and their family, rather than focused on the ownership of who owns that”.
Next steps
This study identifies a gap in understanding neurodiversity, vulnerability and online radicalisation.
Practitioners often generalised connections between autistic traits and extremist behaviours, relying on medicalised and stereotypical perspectives. Such approaches risk criminalising neurodivergent traits rather than addressing systemic factors. The tendency within policing to seek cause-and-effect explanations fosters reductive understandings, which could be problematic when considering the connections practitioners make between neurodivergence and extremism. Drawing parallels with past CT practices focused on Muslim communities, this study illustrates how overemphasis on specific groups distorts understanding and exacerbates stigma. Therefore, the focus on autistic people could influence CT practices in ways that may harm rather than help. Currently, practitioners are missing opportunities to implement effective and bespoke interventions to prevent further harm.
Reframing the narrative of neurodivergence in CT policing
While this study acknowledges algorithm-driven radicalisation risks, it also emphasises the need to reframe how neurodivergence is perceived within CT policing, moving beyond rigid trait-based assumptions like “fixations” and social challenges. These narratives echo stereotypes previously imposed on other marginalised groups and fail to consider systemic barriers faced by neurodivergent people. Instead, CT strategies should adopt a needs-led approach, recognising vulnerability in context. Collaboration with education and mental health services can provide a holistic perspective, reducing both radicalisation risks and criminalisation.
The future of online algorithmic radicalisation
Practitioners agreed that online radicalisation driven by algorithms will intensify, requiring better resources and training. However, such work should avoid underpinning neurodivergent traits into extremist profiles. Tailored neurodiversity training can improve understanding of vulnerabilities without reinforcing stereotypes. Educating young people, particularly neurodivergent people, is equally important. For example, schools could run campaigns highlighting the harms of algorithmic online exploitation. Policymakers and technology companies must collaborate to mitigate algorithmic risks, possibly using AI tools to detect patterns of radicalisation. Ultimately, addressing these issues requires change, both within local CT policing and beyond, into the socio-technological world.
Toolkit and resources for practitioners
To address training costs and time constraints, developing an online toolkit could provide a cost-effective solution. Based on this research, this toolkit would raise awareness, debunk stereotypes and address misconceptions linking neurodivergence with extremism, together with addressing relevant vulnerability factors. Targeting police, educators, social services and healthcare providers, it would feature interactive resources including infographics, case studies, videos and podcasts, with modules on neurodiversity, vulnerability and best practices for collaboration. Further research should evaluate the effectiveness of such training and toolkits in CT contexts.
Contact
- Translational Fellow and author: Dr Alice Siberry, Specialist Criminal Justice Neurodiversity Consultant, Creased Puddle, alice@creasedpuddle.co.uk.
The support of the Economic and Social Research Council (ESRC) is gratefully acknowledged. Grant reference number: ES/W002248/1.