Tenancy fraud isn’t new to the UK lettings market. But when referee-flagged fraud rises by 78.43% in a single year, you pause and ask what's actually going on here.
The reality is that criminals are no longer relying on isolated fake documents. Agencies are increasingly up against organised networks, compromised insider accounts, and real-time AI voice tools built to bypass traditional callbacks entirely.
So how do you verify a reference when you can no longer be sure who, or what, is on the other end of the line?
In this blog, we'll explore how these fraud rings actually operate, why AI voice technology threatens the one channel agencies still trust most, and what a tenant referencing process built to withstand both looks like.
If an agency is still relying on a quick phone call to a referee or a visual scan of a payslip, they’re playing into fraudsters’ hands. Here‘s why:
Today, fraud looks less like a single chancer trying their luck, and more like an organised operation.
As Nishma Parekh, Goodlord's Director of Referencing, explains:
“Rather than one applicant acting alone, there may be several people involved, each supporting a different part of the application. That could include someone answering an employment reference, providing access to an email account, or allowing their company details to be used.”
Within these networks, roles are divided to maximise the illusion of legitimacy. One person takes the call as a ‘former landlord,’ another replies to an email as an ‘employer,’ and a third simply lends a real company's name to the scheme.
Even if an agent questions one thread, the rest of the syndicate can produce corroborating evidence from different angles, because no single person in the chain is carrying the whole lie.
Historically, fake employment references were easier to spot, as either the company didn’t exist, or the phone line was dead. But this has now become the fastest-growing type of tenancy fraud, up by a staggering 226.6% year-on-year.
The reason today’s version is hardest to spot is that the employer is completely real.
Mary Waterfield, Goodlord's Financial Crimes Manager, has watched the shift happen:
“Previously, we saw a lot of fraudulent income documents, provided by individuals declaring fake companies... Recently, we have seen an increase in false employment references being provided by individuals within large and well-known companies.”
That's precisely what makes this so hard to flag on a standard check. The domain is genuine. The company is genuine and recognisable. Only the claim inside it is false. A familiar company name is usually enough to earn trust, which is exactly why these ‘Trojan horse’ references sail through manual checks unquestioned.
Criminals are increasingly using AI to close the gaps that used to give a fraudulent application away, generating mathematically flawless payslips and bank statements in minutes. The more serious escalation, though, is happening over the phone.
The rapid rise of highly accessible generative AI has made synthetic voice cloning far easier to pull off. Where fraudsters once needed specialist technical skills and hours of sample audio, AI can replicate a believable voice almost instantly from a fraction of raw sound.
Goodlord’s Head of Referencing Operations, Nicola Harding, has issued a direct warning:
“Fraudsters could provide someone else's contact details and then use AI voice technology to impersonate them during a live callback.”
Without a robust defence, fraudsters can easily slip through the cracks. The entire purpose of a phone callback is to test the referee with an off-script question and see if they stumble. A real-time LLM-backed voice AI agent can handle that with ease. It pauses, uses conversational filler words, adjusts to unexpected queries, and reacts the way a genuine person would.
Together, AI-generated documents and AI voice spoofing are making the whole fabricated story convincing enough to survive the checks most referencing processes were built around.
Previously, fraud used to carry more risk for the fraudster than the reward justified. AI has flipped that. Generating bank statements and payslips now takes a fraudster minutes and costs next to nothing.
For the agency on the other end, approving that identity costs an average of £9,601, covering legal costs, court fees, rent arrears, void period, and property damage, according to Goodlord's report. That asymmetry is precisely why fraud volumes keep climbing.
The financial sting, however, is only half the problem.
With Section 21 gone, removing a fraudulent tenant means building a case under Section 8, and those proceedings already run to 15 months or more given court backlogs.
The worst part? During those 15 months, the criminals aren’t just living there rent-free; they can actively monetise the property. Once the keys are handed over, the property can be carved up into an illegal HMO, flipped onto short-term holiday let platforms, or used as a ghost address to register further synthetic identities.
By the time the case is finally resolved, the landlord isn't just dealing with lost rent. They're untangling a compliance mess, repairing property damage, and working to rebuild trust with a landlord who was let down by the process meant to protect them.
If the weak point is a human-controlled channel, the solution is not to train negotiators to listen more carefully. It's backing their judgement with verification that doesn't depend on trusting a voice or an inbox in the first place.
To stay ahead of modern fraud networks, the lettings industry needs to lean on Trusted Sources: verifying income, employment, and rental history against immutable data, rather than relying on an email or a phone call that a fraudster is waiting to answer.
The results back this up. Trusted Sources usage has grown from 45% of references in 2024 to over 60% by 2026. Goodlord's Trusted Sources have already helped identify over 97% of fake employment reference fraud, 98% of forged payslips, 83% of referee fraud, and over 80% of fake document fraud on its platform.
That said, a failed Trusted Source check isn't automatic proof of fraud, and none of this replaces experienced staff. The goal isn't removing judgement from referencing. It's giving something more reliable to work from than a voice on a phone.
As Greg Tsuman puts it:
“Technology alone isn't the answer. Digital verification strengthens the process, but experienced people remain essential for interpreting evidence, investigating anomalies, and making informed decisions.”
Even with Trusted Sources in place, it's worth stress-testing if your agency’s current procedures are sufficient to withstand a coordinated AI attack:
Tenancy fraud is no longer a one-person job. It's now a coordinated, AI-assisted network with multiple layers, each one built to survive the manual, channel-based checks that many letting agencies still rely on.
Recognising that shift is the first step to closing the gap. Listening harder on a phone call or scrutinising a PDF more closely won't beat a fabricated story built to survive exactly that scrutiny. The more reliable defence is removing the human-controlled channel from the decision altogether.
This is what Goodlord's Trusted Sources are built for. By verifying income and employment directly through Open Banking and HMRC integrations, it grounds your referencing decisions in immutable data.
Ready to strengthen your referencing process? Get in touch with our team. Or, for a complete breakdown of how fraud rings and AI are reshaping tenancy fraud, download the full Goodlord Tenancy Fraud Report.