5000 Criminals, 28,000 Thousand Tags, And Nobody Watching

The releases were meant to begin on September 2nd. Andy Burnham, three weeks into the job, paused them, ordered a fortnight's review under his new Justice Secretary Alex Norris, and on August 3rd restarted the whole thing for October 1st with rape, serious child sexual offences, and grooming carved out.

Roughly five thousand criminals will now leave early over ten monthly tranches instead of six thousand. Victims' groups got a month and a thousand exclusions. The prison estate got nothing, because the basic arithmetic has not changed.

What has been almost entirely absent from the argument is the receiving end.

Every one of those five thousand men leaves a building where the state knows precisely where he is, and enters a supervision system where the state is currently unable to establish, with any confidence whatsoever, how many of the people it is supposed to be electronically monitoring are actually being monitored.

That is not just going after the state as usual to reinforce the point. It is the alarming position set out by the National Audit Office in July.

Nobody Knows How Many Tags Are Monitored

The tagged population in England and Wales has more than doubled in five years, from around 13,400 in January 2021 to 28,700 by March 2026. Government plans envisage up to 22,000 additional tagged individuals every year from 2027, with demand climbing sharply from this September.

The National Audit Office's verdict on the system about to absorb all of them, delivered in July: electronic monitoring is not working as intended and creates serious public protection risks.

As of March 2026, HMPPS was working through roughly 8,900 cases, some 24 per cent of everyone supposedly required to wear a tag, in order to establish how many were genuinely unmonitored.

Some of those records will turn out to be duplicates, errors, or closed cases.

Nobody should pretend 8,900 offenders were roaming free with a phantom ankle bracelet. The point is duller and worse: the department had to open an investigation into a quarter of its own caseload to find out whether the monitoring was happening at all.

Between August 2024 and July 2025, contractor performance was poor enough that tags were not fitted promptly and potential breaches were not always passed on promptly.

The backlog of fitting, checking, and removal visits peaked at about 7,000 in October 2024.

Performance then improved, and by February 2026 Serco was meeting its timeliness target for fitting visits. It was also, in the same month, managing to get a tag onto a person within two attempts in only 62 per cent of cases.

Those two facts sit side by side in the same audit, and between them they contain the entire British contracting model.

1. The contractor turned up on time.
2. The contractor was measured on turning up on time.

Whether anybody ended up wearing a tag was a separate question, asked by a separate metric, and the answer was no, worryingly often.

Under the Serco contract, details of a breach are not required to reach officials for between 29 and 53 hours. Somebody subject to a curfew can therefore break it on Friday evening and remain, as far as the responsible officer is concerned, a model of compliance until Sunday.

A tag does not stop anybody doing anything.

It is a sensor. It generates information.

Protection of the public depends on a chain of seven separate things going right:

1. The tag is fitted;
2. It works;
3. The data transmits;
4. A human being sees the alert;
5. The right agency receives it;
6. The agency assesses it, and;
7. Somebody has the time to act.

Ministers talk as though the first link is the whole chain.

Every audit of the last two years says the later links are where it snaps.

Probation Has Collapsed

Electronic monitoring does not replace probation officers. It manufactures work for them, and hands it over as an alert. The NAO found probation officers are frequently unable to respond promptly to monitoring breaches because they lack either the information or the capacity.

Between 2023 and 2025, 31 per cent of target probation appointments with offenders did not take place because workloads were unmanageable.

Almost one appointment in three, judged necessary by the service itself, simply did not happen.

The NAO, again, concluded in October 2025 the Probation Service was unsustainable in its existing form without significant corrective action. HMPPS modelling had catastrophically underestimated how long sentence management actually takes; revised analysis indicated the service had been running on roughly half the staff it needed for that work.

Not a shortfall. Half.

September 2026 is when tagging demand is scheduled to rise steeply. The forecast staffing gap in the month is around 1,500 officers. The expansion has been timed to arrive fifteen hundred people early.

The Chief Inspector of Probation put it better than any of this in his 2026 annual report: the service has too few staff, with too little experience, managing too many cases. He is so worried about the quality of risk assessment he suspended the ordinary inspection programme to look at nothing except public protection.

Over 2024 and 2025, two thirds of the probation units inspected were rated inadequate across every area of service delivery. In 2024, practitioners assessed risk of serious harm adequately in just 28 per cent of cases, against 60 per cent in 2018-19.

None of them are activists. They are the government's auditor and the government's inspectorate, describing the government's own service.

Almost half of front-line prison staff, 47.8 per cent, now have three years' service or less, and around half the officer workforce is under forty.

Charlie Taylor, the Chief Inspector of Prisons, has described how established prisoners work on a new female officer, arranging for a couple of inmates to bully her before a charming third party arrives to rescue her and begin the grooming.

Disciplinary cases involving inappropriate relationships between officers and prisoners have risen from twelve in 2017-18 to thirty-two in 2024-25.

A justice system staffed overwhelmingly by people in their first few years is the same system now being asked to manage the most complicated caseload it has ever carried, in the community, at distance, through a contractor's software.

Faced with a capacity shortage, HMPPS has already responded by reducing rehabilitative activity and supervision. Probation absorbed the pressure by doing less probation.

Ministers now propose to relieve prison overcrowding by sending more people to a service which has coped with its existing workload by thinning out what it does.

The alternative to custody is being weakened in direct proportion to how heavily it is used as an alternative to custody.

Unlimited Alarms With No Triage

The NAO found large volumes of alerts and breach notifications generate significant work, while nearly half of recorded outcomes ended in no further action. Serious and minor breaches often enter much the same process at much the same speed.

A device capable of producing an infinite number of notifications does not produce a single additional probation officer.

Where the technology was supposed to multiply supervisory reach, badly designed monitoring instead converts scarce officer time into a need for administrative triage, which is precisely the resource which was already missing. Tagging amplifies the existing problem.

The argument cannot be against tagging, which is a perfectly sensible alternative to a cell for a great many offenders. As far as we know, anyway.

It is against buying a firehose and calling it a tap.

Recall: The Prisoners Come Back

Probation does not only supervise. It recalls. HM Inspectorate of Probation reports recall prisoners now occupy more than 12,500 prison places, at a cost of over £600 million a year.

Many of those recalls are entirely correct, and the officers who made them protected somebody. Recall is frequently the system working exactly as designed, and anybody arguing otherwise should be asked which specific offender they would have preferred to leave at large.

Every early release enlarges the population which is liable to recall. Every stretched supervision arrangement raises the probability of breach within that population. Twelve and a half thousand cells are already filled by people who were released once.

The Ministry of Justice is, at considerable expense, running a machine which produces prison places at one end and consumes them at the other, and has never published an honest net figure.

Ministry of Justice figures obtained by the Telegraph this week supply it, and its ugly.

Recalls to custody over the past twelve months reached 51,419, the highest number ever recorded.

More than 43,000 prisoners have already been let out at the 40 per cent point, and around 20,000 more offenders have been spared custody altogether since the presumption against sentences under twelve months arrived in March.

The men's prison estate is nonetheless running at 98 per cent, holding 83,036 against a capacity of 84,819. Spare places doubled from 1,646 in March to 3,433 by the end of April, then fell back to 1,783, which is roughly where they started. The Institute for Government identifies the mechanism in simple terms: a good many people given suspended sentences breach their probation terms and end up inside anyway.

At least 1,800 cells are out of commission awaiting repairs and fire safety work, and officials are said to be considering delaying that work to free up space. The number of unusable cells in the English and Welsh prison estate currently exceeds the number of empty ones.

A department in that position is being asked to absorb the recall traffic generated by five thousand early releases, on the strength of monitoring its own auditor has condemned.

Zero Funding To Help The Police Cope

The Ministry of Justice estimates the tagging expansion will hand police roughly 4,200 additional bail breaches a year which result in custody. The National Police Chiefs' Council estimated the cost at about £20 million annually.

No specific funding has been provided.

Solving a capacity problem in one department by exporting it, unfunded, to another is such a fixture of British administration it barely registers as news.

No resource disappears.

It changes departmental owner, acquires a new line in somebody else's accounts, and reappears eighteen months later as a different crisis with a different minister apologising for it.

30% Released Into Homelessness

A cell and a bed sit in different spreadsheets. Operationally they are the same problem. In the year to March 2026, 83.3 per cent of prison leavers were housed on the night after release, down from 84.0 per cent the previous year, and part of a decline running since 2021/22.

Roughly one in six people leaving custody has nowhere settled to go on the first night.

The number of cases in scope has risen 29 per cent since 2019/20, to 95,490. Nacro's reading of the 2024/25 figures found around a thousand people a month released into homelessness, a rise of 39 per cent in a year.

Peers debating the issue in July cited a proven reoffending rate of 76 per cent among those released without accommodation, against 36.4 per cent for those released into settled housing.

His Majesty's Inspectorate of Prisons has found establishments still releasing between a fifth and a third of prisoners without a settled address, and independent monitoring boards have called release preparation alarmingly inadequate.

A licence condition makes demands of a homeless man which are close to comic:

• Attend appointments at a fixed address he does not have.
• Keep a curfew at a property which is not his.
• Keep a monitoring device charged when he has no reliable access to a socket.
• Find work while sleeping in a hostel or a doorway.

Then breach, and go back inside, at which point the Ministry of Justice records a recall rather than a housing failure, because recall is the only outcome its systems are built to count.

Renaming Is Cheaper Than Fixing

Probation has spent a decade being restructured. Chris Grayling split it in 2014 under Transforming Rehabilitation, handing much of the work to Community Rehabilitation Companies.

The NAO judged the result poor value for money, the contracts were terminated early, and the service was reunified under public control in June 2021.

Four years on, the same auditor finds deteriorating performance, thousands of missing staff, and unsustainable workloads.

The organisational chart has been redrawn three times.

The underlying requirement has not changed once.

An experienced officer must know enough about an offender, see him often enough, notice deterioration early enough, and have enough hours in the week to act on it.

No managerial technocratic reorganisation has ever produced that, because it is bought with communities and salaries and years, not with a dashboard.

Cooking The Books By Moving The Problem

Government thinks in departmental capacity. Prisons hold X. Probation employs Y. Police number Z. Housing belongs to councils. Monitoring belongs to a contractor. An offender, inconveniently, passes through all of them.

There is one criminal justice capacity system, which the state has chopped into separate budgets and separate accountability.

When ministers announce they will reduce pressure on prisons by supervising more offenders in the community, they are not reducing the burden on the state. They are moving it.

Where probation, police, housing, and monitoring have slack, that is rational and cheaper. Where all four are already at or beyond their limits, the transfer relocates the bottleneck and adds a delay before anyone notices.

Custody is expensive because it concentrates control in one building. Community supervision is cheaper because it distributes control across half a dozen agencies, and it only works if every distributed component functions.

Britain is now attempting the distributed version with:

1. S monitoring system its auditor says is broken;
2. A probation service its inspectorate says is inadequate;
3. A police service which has been handed the bill unfunded, and;
4. A housing pipeline which fails one leaver in six on the first night.

Reoffending already costs an estimated £21 billion a year, which is the price of getting this wrong, itemised in advance.

Burnham's announcement, meanwhile, promised probation officers new powers to ban offenders from driving, from pubs, and from community events, alongside restriction zones agreed with victims.

A service delivering two thirds of the appointments it has already scheduled is to be handed a pub licensing function.

Somebody, presumably, will be responsible for checking. There is no proposal for who.

Fourteen thousand new prison places are promised by 2031. Construction takes years, and prisoners arrive daily.

Alex Norris is meanwhile hunting for room by converting parts of the women's and youth estates into male accommodation, deporting more foreign prisoners, and releasing people held on indefinite sentences which should never have existed.

Each of those is defensible on its own terms. Together they are the improvisations of a department with a fortnight's slack and a fixed appointment on October 1st.

Between now and 2031 the state proposes to hold the line with technology it cannot audit, officers it has not hired, accommodation it has not built, and a police force it has not paid.

The prison crisis has not been solved by releasing more offenders. It has undergone an accounting trick which will also fail when the book-cooking becomes subordinate to the levels of crime being reported at police stations by ordinary people.

It has been distributed across probation, monitoring, policing, and housing, and when those fail it returns through the same gate it left, wearing a recall notice.