O2 Rolls Out Free AI Scam Call Blocking as 61% of Brits Report Suspicious Calls

O2 has rolled out free, AI–powered scam call protection to all of its customers, following a successful trial earlier this autumn. The service, built on technology from O2’s existing partner Hiya, is expected to block more than 3 million malicious calls a week in real time — stopping them before the phone even rings. Customers do not need to do anything: the protection is applied automatically across O2’s network to all compatible devices.

The timing reflects a stubborn problem. Virgin Media O2’s latest quarterly Scam Index found that 61% of Brits who encountered a suspected scam between July and September 2026 said it involved a suspicious phone call, up five percentage points on the previous quarter. Text–message filtering and email security have improved, but voice remains one of the most exploited channels for reaching people — including employees at UK businesses.

For organisations, the announcement is welcome news and a useful prompt. Network–level blocking will stop a large volume of junk before it reaches staff mobiles on O2. It will not stop every call, it does not cover every line a business uses, and it cannot stop a convincing caller who gets through from talking someone into a payment, a password reset or a change of bank details.

From flagging to blocking

O2 has been identifying suspicious calls for some time. Since the start of 2024 it has flagged more than 1.6 billion suspected scam and spam calls. Some are blocked using Ofcom’s “Do Not Originate” list — numbers that should never make outbound calls, such as certain numbers used only to receive calls. The new system significantly expands the range of fraudulent calls that can be blocked outright, rather than merely labelled.

Murray Mackenzie, Virgin Media O2’s Director of Fraud Prevention, described the change as “evolving our AI scam call detection service from flagging to blocking fraudulent callers, stopping that first ring in real time”. He urged customers to keep reporting new scam numbers to 7726, the free reporting service that feeds the system.

The distinction matters. A warning label on an incoming call still relies on the person answering to notice it and act on it. A blocked call never reaches them at all. For a busy employee taking calls on a work mobile, the difference between “Suspected spam” on the screen and silence can be the difference between a fraud attempt and no attempt.

How the system works

According to O2, the AI–assisted system monitors the network for nuisance callers and continuously refines its detection. It draws on two main sources: customer reports sent free to 7726, and ongoing analysis of call behaviour. That makes it different from older approaches that depend solely on static blocklists of known bad numbers.

Static lists struggle against scammers who rotate numbers quickly. By the time a number is reported, confirmed and added to a list, the caller may have moved on. Behavioural analysis aims to spot the pattern of a scam operation — the volume, timing and spread of its calls — and act on it even when the specific number is new. Customer reports then help confirm and sharpen what the system has seen.

O2 also uses firewall technology for text messages, which it says has prevented more than 1.4 billion fraudulent texts from reaching customers to date. The new call protection brings a comparable network–level approach to voice.

What the trial revealed about scammers

The trial data offers an unusual glimpse into how scam operations behave when they are blocked. O2 found that 60% of blocked calls were followed by a repeat attempt within a minute, with scammers often unaware they had been blocked. On average, a number flagged as fraudulent attempted three times as many calls after being blocked as before.

That tells us two things. First, many scam operations are automated or high–volume, dialling repeatedly rather than carefully selecting targets. Second, blocking does more than filter: it wastes scammers’ time and resources. Every call that fails silently is a call that did not reach a potential victim, and every retry is effort spent on nothing.

The five–point rise in a single quarter is notable. It suggests that, as other channels become harder to exploit, scammers are leaning more heavily on voice. A phone call carries a sense of urgency and personal contact that a text or email often lacks, and a skilled caller can adapt in real time to the person on the other end.

The timeline so far

The sequence below is drawn from O2’s announcement and the accompanying reporting. Where no specific date was given, the step is described rather than dated.

The policy backdrop

Fraud minister Lord Hanson welcomed the roll–out as part of the government’s Fraud Strategy and Telecoms Charter, under which mobile networks are working to block scam calls before they reach the public. The charter approach places expectations on the networks themselves to act at scale, rather than leaving individuals to spot and avoid every scam.

That makes sense. Networks see patterns no individual can: the same caller dialling thousands of numbers, unusual bursts of activity, numbers that should never originate calls. Blocking at that level protects everyone on the network, including the people least likely to recognise a scam.

Ofcom has also been scrutinising how numbers are misused. Our report on Ofcom’s investigation into VoIP numbers used in scams looked at how internet telephony can be abused to reach victims, and why providers are under pressure to tighten controls.

Why this matters for UK businesses

Scam calls are not only a consumer problem. Businesses are attractive targets because their staff can authorise payments, reset accounts and grant access. A phone call is often the opening move in attacks that end in serious losses.

Common business–targeted voice scams include:

• Supplier and payment fraud. A caller posing as a supplier asks finance staff to update bank details before the next invoice is paid.
• Executive impersonation. Someone claiming to be a director, often “travelling” or “in a meeting”, pressures a junior colleague to make an urgent transfer.
• IT support impersonation. A caller posing as the help desk or a software provider asks for a password, a one–time code or remote access to fix a “problem”.
• Bank and HMRC impersonation. Callers claim to be from the business’s bank or a government body and demand immediate action.
• Help desk manipulation. Attackers call an organisation’s own IT support pretending to be a member of staff, trying to get a password or multi–factor authentication reset.

Network–level blocking helps with the high–volume end of this spectrum: robocalls, mass–dialled scams and numbers already flagged as fraudulent. It is less effective against targeted calls from fresh numbers by someone who has researched the business. Those calls are fewer, but they tend to be the ones that cause the largest losses. Our coverage of the BT email password reset attack shows how account recovery processes can become the weak point.

Layers of protection against voice fraud

The chart below is our indicative assessment — not measured data — of how much each layer reduces the risk of a smaller UK business losing money or access through a scam phone call.

The pattern is clear. The most effective controls are process controls that make a successful call harmless: verifying bank detail changes through a known number, requiring two people to approve large payments and checking identity before resetting credentials. Network blocking is valuable because it reduces the volume of attempts, but it works best alongside those processes, not instead of them.

The bottom bar matters most of all. Caller ID can be spoofed, so a call that appears to come from your bank, a supplier or a colleague proves nothing on its own.

A readiness score card for business phone security

The score card below rates common positions for UK businesses. It is a judgement framework for discussion with your IT and telephony providers, not an audit result.

The last row surprises many businesses. After years of adding mobiles, moving to hosted VoIP, adopting Teams calling and keeping an old landline “just in case”, few organisations have a single list of every number they own, who answers it and what protection it has. With the PSTN switch–off approaching, as our report on the final months before the switch–off explains, that inventory is worth building anyway.

What stronger voice fraud defences cost

The figures below are indicative ranges for a smaller UK business with up to around 50 staff. They are planning estimates rather than quotations, and they vary with the number of users, sites and phone services involved.

The cheapest controls are often the most effective. A rule that bank details are only ever changed after calling the supplier back on a number already on file costs nothing to introduce and defeats one of the most common and damaging scams. A single successful payment diversion can easily cost more than all of the measures above combined.

Network blocking versus business process controls

The two columns are complementary. Network blocking reduces how often staff are tested; process controls decide what happens when a call gets through. Businesses that rely on one without the other are leaving a gap.

How ready are smaller UK businesses?

The gauge below is our indicative estimate of how well a typical smaller UK business is protected against losses from scam phone calls. It is a judgement based on the patterns in the score card, not a survey figure.

A score of 35 reflects a familiar picture. Email filtering is usually in place and staff have had some phishing training, but phone calls are treated as trustworthy by default. Payment changes are sometimes verified, sometimes not, depending on who takes the call. The help desk, often a small team or an outsourced provider, may reset passwords for anyone who sounds plausible. O2’s blocking will nudge that score upwards for staff on its network, but the biggest gains come from tightening processes.

Analysis: a welcome step that shifts the battleground

O2’s move is a genuine improvement. Blocking more than 3 million malicious calls a week before they ring removes a great deal of noise from people’s lives, and the trial data shows it frustrates scam operations in a measurable way. The continuous learning from 7726 reports and call behaviour is a sensible response to scammers who change numbers faster than static lists can keep up.

But scammers adapt. If high–volume, automated calling becomes less profitable on the major networks, some operations will move towards fewer, better–targeted calls, other channels, or lines and services the network filters do not cover. The rise in phone–based scams recorded in VMO2’s Scam Index suggests the contest is far from over.

For businesses, the practical conclusion is that network protection should be treated as a bonus, not a strategy. It lowers the number of times staff are tested. It does not decide whether they pass. That depends on processes, culture and the systems behind the phone: who can change a payee, how the help desk verifies a caller, whether a director will back an employee who refuses an urgent request.

There is also a telephony angle. Many organisations are moving from traditional lines to hosted VoIP and Teams calling as the analogue network is retired. That migration is a chance to choose providers and configurations with fraud controls built in — call filtering, blocking of suspicious international destinations, alerts on unusual outbound activity and clear reporting. Our coverage of Vodafone’s copper switch–off and VMO2’s cost cuts and business continuity sets out other questions worth asking during that move.

At a glance

For related coverage, see our reports on Ofcom’s investigation into VoIP scam numbers, the PSTN switch–off countdown, Ofcom’s view of AI in telecoms customer service, Vodafone’s copper switch–off and the BT email password reset attack.

Frequently asked questions

O2’s blocking will quietly stop millions of scam calls before anyone has to decide whether to answer. For businesses, the calls that still get through are the ones that matter — and the answer to those lies in processes and people, not filters alone.

Cloudswitched

London-based managed IT services provider offering support, cloud solutions and cybersecurity for SMEs.