When practitioners talk about online radicalisation, they often describe it as something deliberate: searching, seeking, choosing. Findings from this research suggest something more unsettling: that for many young people, the journey into extremist spaces is not intentional, but incremental, normalised, and shaped by algorithms long before anyone realises where the line is.
This matters, because risk does not always arrive looking like risk.
The online world as the trigger
Across all interviews, practitioners described the online world itself as the key trigger. Engagement rarely began with overtly extremist material. Instead, it started with content that was familiar, social, and seemingly harmless, such as gaming, humour, memes, or identity‑based content shared within online communities.
From there, individuals were exposed to increasingly adjacent material. Algorithms played a central role, responding to clicks, views, and engagement by providing content that was slightly more provocative, slightly more extreme, and designed to hold attention. Several practitioners described this exposure as passive: young people did not actively search for extreme content, they simply encountered it.
Once exposure begins, choice becomes blurred. Participants spoke about individuals being presented with “paths”: some safer, some riskier. Which path was followed often appeared down to chance rather than intent.
Going “down the rabbit hole”
Every participant used the same language to describe the next stage: going down the rabbit hole.
This was the point at which content stopped shocking and started feeling normal. Material that once would have provoked discomfort no longer did. Graphic imagery, racist language, extremist narratives, and harmful memes became familiar currency within online spaces.
Participants described a process of desensitisation, where escalation became necessary to provoke reaction, either internally, or from others online. At this stage, content becomes more extreme not because of ideology, but because the previous level no longer exists.
For some, this transition was described as active, with participants describing a person “stepping over a line”. For others, it was passive’ “falling foul”, “being pulled in”, or “being drawn down”. What was consistent was the outcome: extreme content became mainstream within their online world.
When online normality collides with the offline world
Practitioners expressed particular concern about what happens when online engagement no longer produces enough response or stimulus. At this point, risk shifts.
Some participants described a movement from online behaviour to what they termed “finishing” or “concluding” actions, where offline behaviour is used to recreate the reactions previously achieved online. This is where extremist engagement becomes visible to systems, and often where enforcement begins.
Yet what concerned practitioners most was not just the behaviour, but the difficulty in identifying where the process began. In many cases, there was no clear origin point, grievance, or ideological entry. There was simply a pattern of engagement reinforced by algorithms.
Is this about neurodivergence or about being human online?
Although practitioners frequently framed this process as being heightened for neurodivergent people, particularly autistic individuals, the mechanics described were not unique. The need for reaction, stimulation, and escalation appears consistent across neurodivergent and neurotypical populations.
- This raises an important question for practice:
Are we sometimes attributing risk to neurodivergence, when the real driver is how online environments function?
If algorithms reward escalation, boundary‑pushing, and emotional engagement, then the “pull” they create is not a niche problem. It is structural and it can affect anyone.
Reflections for practice
This finding invites practitioners to pause and ask:
- At what point does harmful content become normalised for the individual?
- Are we intervening too late, once behaviour has already escalated offline?
- Do our assessments take account of algorithmic exposure, or focus only on personal traits?
- And how clearly do we distinguish between curiosity, desensitisation, and ideology?
Understanding how extreme content becomes mainstream is essential, because once it feels normal, recognising risk becomes much harder.
This blog is part of a five-part series exploring how online algorithms, neurodivergence, and systems of response intersect in practice, based on Dr Alice Siberry’s research. To continue the series and access further practitioner resources, sign up here.
If you would like to explore how we could help improve understanding of neurodiversity in your organisation, please get in touch and we would be happy to discuss this with you.