When you click on a retailer link on our site, we may earn affiliate commission to help fund our not-for-profit mission. Find out more.
Some crime reports made by scam victims are never being read by actual police staff, Which? believes.
Fraud victims in England, Wales and Northern Ireland are urged to report crimes straight to a centralised reporting centre called Action Fraud, via phone or web form.
But reports are assessed by artificial intelligence in the first instance, and the City of London Police (which runs Action Fraud) has repeatedly failed to guarantee that every report is then read by its crime reviewers.
It comes after a four-month investigation into Action Fraud, which also found:
- a senior police source branding the system ‘not fit for purpose’
- claims that a solvable case was mysteriously denied an investigation
- Action Fraud screening out reports where flustered and panicked victims have omitted some details
- local police forces being given ‘appallingly’ low numbers of detailed fraud cases to investigate.
This comes just weeks after an undercover investigation by The Times newspaper discovered poor treatment of fraud victims by a firm contracted by the City of London Police to deal with cases. Victims were ‘mocked’ by staff and misled about investigations into their issues.
Read our full investigation: you can read our deep-dive into fraud in the latest edition of Which? magazine.
Crime victims screened by an algorithm
In 2018, a Which? survey found that 82% of respondents thought their local police station was the appropriate place to report fraud.
We hear from dozens of scams victims every week. They tell us they expect to call or visit their local station, before a visit from a local officer, so they can hand over crucial evidence.
They believe that the crime will be investigated, the money traced through the banking system and the perpetrator brought to justice.
Yet the vast majority of victims don’t receive a local, personalised response.
If you try to reach your local force about a fraud or scam by calling 101, in most cases, unless there’s an immediately identifiable local suspect or it’s a crime in action which elicits an immediate ‘call for service’, you will simply be told to call Action Fraud (or fill in its web form).
That is, unless you are in Scotland, which recently pulled out of the Action Fraud system in favour of maintaining its own database.
So what does Action Fraud do with your report? As a centralised fraud and cybercrime reporting centre, it has no investigative powers. Reports are controversially screened and scored by artificial intelligence in the first instance, based on numerous factors including risk, vulnerability and age of the victim.
Some fall at the first hurdle; others are passed to crime reviewers (trained police staff) at the National Fraud Intelligence Bureau (NFIB) for consideration. In theory, the reviewers identify the reports most suitable for investigation.
The chosen reports are then sent to whichever local police force is most connected to the case (for example, the force where the suspect lives), and it’s up to local officers whether to pursue an investigation.
Some forces aim to investigate all frauds referred in this way, while others reject the majority because they lack resources or don’t see it as a priority.
Most victims who contact Action Fraud are unaware of the gauntlet their report must run to even reach an investigating officer’s desk at a local force, or the postcode lottery that it faces if it does get there.
Panicked victims may be overlooked
Which? travelled around England interviewing senior police officers at City of London Police (which runs Action Fraud and the NFIB), local forces, Trading Standards, crime prevention experts, victims and those who support them in the aftermath of fraud.
Our first port of call was City of London Police, the national lead force for fraud, which runs Action Fraud and the NFIB.
There, we met Commander Karen Baxter, the national co-ordinator for economic crime.
Because fraud reports are assessed by algorithms in the first instance, we asked Baxter whether it’s possible for a victim to submit a report to Action Fraud and it not be seen by a human being at all.
She wasn’t sure and said she would need to check, but assured us the ‘vast majority’ of residual reports are brought through ‘some sort of crime review’ and looked at manually by a crime reviewer, albeit in ‘slower time’.
After the interview, we again asked the City of London Police if it can guarantee that every fraud report is read at some point.
This time, we received a statement confirming that there are ‘a minority of cases where the details are such that it is clear they can’t be progressed, and those may not be reviewed by a person.
‘For instance, where the person reporting has not included what would be considered as viable lines of enquiry, such as details relating to the suspect, bank account details and a description of the circumstances of the offence.’
Action Fraud doesn’t let you submit physical evidence with your phone or web form report, but victims are asked to give details they may have, such as bank account numbers or a suspect’s name. So if you omit details when typing in a panic at losing life savings, your report is less likely to be read.
Find out more: how to report a scam.
Revealed: how the police score fraud victims
The practice of assessing reports by algorithm is one Baxter staunchly defends: ‘With 800,000 contacts a year we’d simply not be able to go through every one.’
She added: ‘If we were to review manually every report that comes in in the first instance, we would miss the risk factors.
‘Our view, and it’s the right view, is we need to identify the highest level of risk as quickly as possible. On occasions, people contact Action Fraud and the offence is going on as they’re reporting it. We need to intervene really quickly, get local police deployed and stop that offending immediately.’
She said there will also be cases of ‘lower risk but significant vulnerability’ that need to be picked out quickly.
‘Velocity of growth’
In a separate interview, NFIB head of fraud, Chief Inspector Paul Carroll, elaborated on the algorithm and explained that reports are scored on the ‘velocity of growth of a case’.
For example: ‘A case may only consist of two reports. Within a day you could suddenly see a huge growth [in the number of reports], turning into quite a big network.
‘The score tells us whether the case is growing rapidly.’
There’s also vulnerability scoring that ‘gives us a feel as to your level of vulnerability – whether you’re a repeat victim, or whether there’s a big effect on the finances, and so on’.
A ‘risk word search’ combs reports for words linked to mental health, suicide or terrorism, and forwards them straight to victim services or Special Branch as appropriate.
Victims under the age of 14 are also singled out for special support. He is adamant that there isn’t anything that automatically separates out victim losses of more than £100,000, after a report last year in The Times claimed the ‘computer says no’ to investigating cases below that threshold, and stressed that some cases under this threshold do progress.
Police officer claims ‘Action Fraud is not fit for purpose’
Inspector A leads a victim services team at a police force in southern England – we are choosing not to name her or the force she works for to protect the inspector and her staff.
Although she was at pains to stress the ‘very good’ working relationship her force has with Action Fraud, and her respect for its staff, she added: ‘I think the general policing thoughts are that it’s not fit for purpose.
‘It’s traditionally been quite difficult for people to get through [on the phone line]. Online reporting [via the web form] isn’t suitable for people who are vulnerable and don’t know a lot about fraud. You have to pick your own classifications for fraud. There are many . I don’t know them all. It’s a lot.’
And their list went on: ‘They don’t disseminate things we really should investigate. They just screen them out. And when I say ‘they’, I mean the organisation, not individuals.
‘And the victim data. Surely if they can send it to us every month they could send it to us every week?’
The comment refers to a spreadsheet sent to their team every month by Action Fraud, with very basic details of any fraud reports sent by local residents – typically around 600 to 650 a month. Since we spoke, Action Fraud has begun sending the data weekly.
Victims have been ‘forgotten’
However, when we asked if Action Fraud is ‘salvageable’, Inspector A conceded that she ‘can’t think of anything any better’.
If forces were required to record and screen all fraud, she fears some wouldn’t cope. Plus, Action Fraud lets people ‘report something suspicious that isn’t actually a crime yet’, giving early opportunities to disrupt scammers.
She has been running a pilot scheme since April 2019 – one she believes is unique among police forces – in which they call every single victim on that spreadsheet to offer practical advice (for example, how to change your password or enable two-factor authentication) and emotional support.
It takes up to a month to ring everybody, and then the next spreadsheet arrives.
So does the system as a whole (Action Fraud, NFIB and allocations to forces) take enough account of victims and their needs? We posed this question to Inspector A.
‘No, I don’t think it takes any account. It’s basically a crime-recording function and was never supposed to take care of victims. That was always supposed to be in-force. I think victims have been forgotten, but they’re being remembered now.’
Inspector A was referring to new national economic crime victim care units, which were set up by the City of London Police last year to better support fraud victims.
Fraud investigations ‘appallingly’ low
In the same building at the same force, we meet Thomas*, a retired police officer who works for a not-for-profit business with a social mission.
The business has teamed up with the police in a public-private partnership to provide victim services, and Thomas speaks to fraud victims on the phone.
He was keen to tell us about a case which he says didn’t pass the Action Fraud algorithm test, despite having undeniably clear lines of inquiry.
In 2018 he spoke with a victim of debit card fraud, whose basic details had come across on the spreadsheet.
They described how thousands of pounds had repeatedly disappeared from their bank account during periods away from home. Thomas’s police training kicked in and he began to suspect someone close to the victim who had access to their home.
Some fraudulent transactions had taken place in local shops, meaning CCTV could identify the perpetrator.
Yet when Inspector A chased up Action Fraud (where the victim originally reported the crime), she found the case hadn’t gone anywhere because it didn’t meet the threshold.
When Thomas encouraged the victim to report the case to the local force, the person he suspected was subsequently arrested.
We don’t know why the system rejected a case with a vulnerable victim and clear investigative parameters. Inspector A told us the NFIB only disseminates in detail around 10 cases a month to her force for investigation – a number she branded ‘appalling’.
No justice for victims
No system is perfect. Before Action Fraud, local police forces often failed to share fraud reports with each other, so scam outfits could target victims around the UK without anyone spotting the bigger picture.
The introduction of Action Fraud had at least one major benefit. Expert crime reviewers and analysts at NFIB receive fraud reports from all over the UK and form a broad intelligence picture. They can spot groups of scams linked by bank account details, names or other data (known as ‘networks’) and act quickly to allocate them to forces, or disrupt them.
The disruptions team has a hotline to major businesses and can request that scam websites, social media accounts, phone numbers and bank accounts associated with fraud are closed.
Yet these benefits will be meaningless to the hundreds of thousands of people who report crimes through the system each year to find that justice is never served.
And because this sort of activity happens behind the scenes, victims – particularly those whose cases go no further – are never made aware of how their report helped to protect others.
After a wait often stretching to months, they’ll receive a letter concluding that no viable ‘line of enquiry’ could be identified.
Find out more: join the Which? campaign to stamp out scams.
The City of London Police responds to Which?
When we presented our findings to the City of London Police, it told us the Action Fraud/NFIB system is ‘the envy of police forces around the world’, adding ‘every report is important to Action Fraud’.
It said every report develops a clearer picture of offending, helping investigate fraudsters. It added that it analyses each report to ensure it responds appropriately. It said crime reviewers sometimes contact victims if pertinent details are missing from their reports.
How to avoid a scam
If you receive a request for money, personal or banking details – whether by phone, letter, online or in person – don’t respond straight away.
Fraudsters will try to rush or panic you into complying, before you have a chance to think clearly. Hang up, shut the door or walk away from the screen and take five minutes to reflect.
Always report fraud to Action Fraud at actionfraud.police.uk or on 0300 123 2040. Doing so can help police see the bigger picture and stop the scammers.
* Some names have been changed to protect the identity of our sources.
Find out more: how to spot a scam.