Jump to...
In the span of one week, two Prevent Learning Reviews have been published: the first in relation to Axel Rudakubana who killed 3 young girls in Southport in 2024; and the second in relation to Ali Harbi Ali who killed the MP Sir David Amess in 2021.
Both attackers had been referred to Prevent, yet neither was stopped before carrying out their deadly acts.
Last week, we outlined five reasons why the Prevent Learning Review related to Rudakubana is misleading.
This week, we turn our attention to the review of Harbi Ali – another report, another reviewer, but similar critical flaws, leading also to misleading conclusions about the future of Prevent.
Prevent’s ‘risk’ assessments continue to miss the mark
Unlike Rudakubana who was referred to Prevent three times, Harbi Ali was referred to Prevent just once, seven years before he killed Sir David Amess.
His referral came when his secondary school, concerned by a sudden decline in his academic performance and his adoption of Islamic dress, suspected he might be radicalised. At that time, Harbi Ali had no prior contact with police or social services.
When interviewed by Prevent, the main issues identified were his views on music and interest on loans, which he considered to be haram (forbidden to Muslims).
For this, he was deemed suitable for Channel, the voluntary deradicalisation programme, and recommended for five sessions to address these concerns.
After completing the sessions, the Home Office-approved independent Channel provider – described as “expert” in extremism and ideology – concluded that no further sessions were necessary. Harbi Ali was judged to be a low counter-terrorism threat.
Prevent’s obsession with ideology distracted from domestic issues
The review notes that some family issues were raised, and the local authority was tasked with assisting him.
However, the reviewer lacked access to the relevant documents to clarify the outcome. What is clear from the final statement by the Channel Panel on 23 April 2015, is that the case was closed and that these family issues had been resolved:
“This case should also be closed. [CTCO] wrote to the intervention provider. AA is back in school. [Certain family issues] have been sorted out. If the intervention provider is happy, AHA will be exited form (sic) Channel.”
The reviewer concludes that these family issues may not have been pursued vigorously enough.
Additionally, the review suggests that today’s Prevent referral process would have addressed other concerns that were not considered at the time—such as Harbi Ali’s views on non-segregation and his desire to live in an Islamic state (not the same thing as ISIS) – though these issues were discussed
However, this assumption is misleading. It presumes that the case was closed without further scrutiny. In fact, it was subjected to six-monthly reviews for a year.
And yet, despite these regular check-ins, Prevent failed to identify any indications of Harbi Ali’s violent intentions in the seven years leading up to the murder.
Prevent fails and communities pay the price
This exposes a deeper issue: even with ongoing oversight, Prevent remains fundamentally incapable of detecting and preventing genuine threats.
Yet, like the review of Southport, this one concludes that more surveillance and more Prevent are the answers.
This is a recurring issue. Despite its sweeping reach—more than 58,000 referrals since 2015, many involving children—Prevent has failed to prevent even the few violent acts it has ostensibly been meant to address.
Out of these tens of thousands, fewer than 10 were later linked to violent offenses. Yet, even in these rare instances, Prevent did nothing to stop them.
The key issue is not that Prevent needs more power, but that it fundamentally fails at its core aim: distinguishing between those who pose a real threat and those who do not.
Even when it identifies someone who might be a risk, as any system with such a broad reach inevitably does, it still cannot prevent harm.
The result is a growing dragnet that wrongfully targets thousands of young people, yet still fails to fulfil its fundamental purpose of protecting society.
It does this while repeatedly describing the 99.9% of referrals where nothing subsequently happened as “successes” as if a causal relationship can be established in a situation when a predicted outcome has not occurred.
Recycled reviews and repeated mistakes
Although the Prevent review for Harbi Ali was commissioned in 2021, it was only published on 12 February 2025, a week after the Southport review.
Due to ongoing inquiries and judicial processes, both reviews are limited, based on desk research, and only drew from the paper trails of Prevent referrals, without direct interviews with those involved.
However, the reviewer in the Harbi Ali case seems oblivious to the limited nature and utility of this review; much like the recommendations from the Southport review, it argues for even more data retention and surveillance in a way that is confused, contradictory and ultimately harmful.
The argument in this review is that if the material in this case had been deleted after five years as is the current requirement, then it would have been “nigh impossible to conduct this review”.
However, this assumes that everyone referred to Prevent, the majority of which never go on to do anything wrong, should remain under surveillance indefinitely.
This approach exposes the fact that Prevent databases are akin to criminal intelligence databases.
It also risks unfairly labelling thousands of innocent individuals – most of them children and young adults – as future threats.
A clear path to expanding surveillance
The timing and nature of these reviews reveal a troubling trend. Against a backdrop of Prevent reviews linked to high-profile murders, the government frames a “new type of terrorism” and calls for more surveillance.
Meanwhile, it is failing to address the root causes of violence.
This is not a genuine review of Prevent. A true review would critically assess its systemic failures and recognise that Prevent has once again failed to prevent harm.
Despite counter-terrorism involvement in both cases – through Vulnerability Support Hubs in Rudakubana’s case and the Channel intervention and ongoing monitoring in Harbi Ali’s case – neither measure prevented them from carrying out their crimes.
Unless we confront the failure of Prevent and the misleading nature of these reviews, we risk continuing down a dangerous path, where political performance trumps genuine security, sacrificing rights and freedoms without making anyone safer.
Related…
The People’s Review of Prevent: A Response to the Shawcross Report
