On 6 August 2026, the UK Home Office released data showing that 10,293 people were referred to a government programme called Prevent in the year to September 2025. That is a 39% rise from 7,408 the previous year, and the highest number since records began 11 years ago The Guardian.

Prevent is a UK government programme that tries to steer people away from extremist views before they commit acts of violence. Think of it as an early-warning system. When someone is flagged as a possible concern — by a teacher, a doctor, or the police — they are referred to Prevent. If the concern is serious enough, the case moves to a more intensive stage called Channel, where people get tailored support.

The referrals covered 9,957 individuals. The surge follows a sharp increase after the July 2024 Southport attack, when Axel Rudakubana killed three girls at a dance class. Rudakubana had been referred to Prevent three times before the attack, but his case was closed each time because no clear ideology could be found. That failure has shaped how the programme is scrutinised today.

The numbers reveal a shifting picture of what kinds of extremism are being flagged. Extreme rightwing referrals rose 20% to 2,103, making up 20% of all cases, up from 1,480 the year before. Islamist extremism referrals fell to 860 (8% of the total), down from 945. Another 818 referrals (8%) involved a fascination with extreme violence or mass casualty attacks, 95 (1%) related to InCel extremism, and 32 (0.3%) to leftwing extremism. A further 353 referrals (3%) involved multiple ideologies with no dominant one.

More than half of all referrals — 56% — were for people with no identified ideology at all. In the previous reporting year ending March 2025, that figure was 34% Reuters. The jump from 34% to 56% reflects a broadening of the programme to capture people showing worrying behaviour even when they are not driven by a specific ideology. This shift intensified after the Southport killings exposed gaps in Prevent's ability to act on people fixated on violence without an ideological motive.

Children and teenagers make up the largest share of those referred. Where age was known, 11-to-15-year-olds were the biggest group at 3,655 referrals (36%), followed by 16-to-17-year-olds at 1,379 (13%). Combined, under-18s accounted for nearly half of all referrals. Of 5,315 referrals where ethnicity was recorded, 3,600 (68%) were white, 949 (18%) Asian, 401 (8%) black, and 365 (7%) listed as "other." Among the 1,100 Channel cases where ethnicity was known, 835 (76%) were white, 159 (14%) Asian, 59 (5%) black, and 47 (4%) "other."

The demographic profile has drawn sharp criticism from civil liberties groups. Nearly one in five referred individuals (18%) were autistic, and 36% had at least one mental health or neurodiversity condition listed on their Prevent record. Jacob Smith, team leader at Rights & Security International, said there is "no excuse for unleashing this much suspicion against autistic people" The Guardian.

Security minister Dan Jarvis framed the figures as evidence of the programme's reach, noting that more than 6,000 people were moved away from violent ideologies through Prevent. He said the record numbers "underline the scale of the challenge in protecting the public from terrorism."

The trend has been climbing across several reporting periods. Home Office data for the year ending March 2025 showed 8,778 referrals (a 27% increase on the prior year) involving 8,517 individuals Gov.uk. Of those, 1,727 were discussed at a Channel panel and 1,472 adopted as Channel cases The Guardian. The jump from 8,778 (March 2025) to 10,293 (September 2025) over just six additional months signals that the post-Southport surge has not levelled off but accelerated.

The broader context here is one of tension at the heart of the programme. The large share of autistic individuals and those with mental health conditions raises the question of whether Prevent is operating as a counter-terrorism tool or as a de facto mental health triage system under a security label. The 76% white share of Channel cases aligns with the 68% white share of overall referrals but diverges sharply from the historical emphasis on Islamist risk, which has typically produced higher proportions of Asian referrals. The 20% increase in extreme rightwing referrals alongside a decline in Islamist referrals points to a measurable rebalancing of the caseload, though the 56% with no ideology at all remains the single largest category by a wide margin.

The Rudakubana case casts the longest shadow over this data. His three Prevent referrals were closed because no dominant ideology could be identified, placing him in the same category that now covers more than half of all referrals. In my view, whether these record numbers mean the system is catching threats earlier — or simply casting a wider net faster than it can properly assess — is the central question the Home Office data does not answer.