Interpol just put a number on the fraud crisis: $442 billion in global losses for 2025 alone. Their Global Financial Fraud Threat Assessment, published in March 2026, calls financial fraud one of the top five global crime threats, right alongside drug trafficking and money laundering.
Most of the coverage has focused on the headline figure. But the Interpol financial fraud report 2026 tells a more specific story if you read past page one: behind the billions are fake invoices, forged credentials, synthetic identity kits and AI tools that produce convincing fakes faster than any human reviewer can spot them.
At VerifyPDF, we see this supply chain up close every day. And it is accelerating.
$442 billion in fraud losses and the documents that fuel them
The $442 billion estimate comes from the Global Anti-Scam Alliance, cited in the Interpol report. Interpol itself notes the real number is likely higher because of significant underreporting. So think of $442 billion as the floor, not the ceiling.
What struck us is not the total. It is what sits underneath it. Interpol’s fraud-related Notices and Diffusions (the international alerts they send to member countries) increased 54% year-over-year from 2024 to 2025.
In the UK alone, fraud now accounts for 43% of all estimated crime, with losses rising 33% to approximately $3.5 billion. In the Americas and Caribbean, fraud surged 40%. Europe accounted for 38% of all fraud-related Notices, followed by Asia-Pacific at 28%.
The report launched at the Global Fraud Summit in Vienna, jointly organized by Interpol and UNODC, with over 1,300 participants. That alone tells you how seriously law enforcement is taking this.
But here is the part that gets lost in the policy discussions: behind almost every fraud category Interpol tracks, there is a document. A forged invoice. A manipulated bank statement. A synthetic identity built from AI-generated credentials. The document is the weapon, and most organizations still do not inspect it properly.
BEC is the world’s most reported fraud type (and it runs on forged invoices)
Business email compromise topped Interpol’s list of reported fraud types, based on operations and investigative support requests from member countries. Most reports came from Asia-Pacific and Europe.
If you are not familiar with BEC, here is how it typically works: criminals compromise or spoof an executive’s email account, then send instructions to employees (or vendors) to wire money to accounts they control. The FBI’s Internet Crime Complaint Center puts BEC losses at $2.77 billion in 2024 alone, with a cumulative $55 billion globally since 2013. The average loss per incident hit $137,000, up 83% from 2019.
Here is what most coverage misses: BEC does not run on emails alone. The actual trigger for the wire transfer is almost always a document. A forged invoice with updated “new” bank details. A fake purchase order. A manipulated contract.
The email is the delivery mechanism, but the fake document is what makes it convincing enough to bypass internal controls. And criminals know this.
Interpol cited a specific case where a Massachusetts-based workers’ union lost $6.4 million in a single BEC scheme. In another case, $7.7 million moved across multiple accounts and jurisdictions within hours.
These are not edge cases. According to Trustpair’s 2026 Fraud Trends survey, 58% of organizations report invoice fraud as their most common attack form.
In our experience, the invoices used in BEC attacks have gotten disturbingly good. They match the formatting, fonts and even metadata patterns of legitimate documents from the same vendor. A finance team member scanning it at 4pm on a Friday is not going to catch it.
As we covered in our analysis of the rising threat of fake bank statements, financial documents have become the favorite weapon for exactly this reason: they look real, they feel real and nobody is checking them at the structural level.
The “industrialization of fraud” is not a metaphor
Interpol Secretary General Valdecy Urquiza used a phrase in the report that deserves more attention than it got: the “industrialisation of fraud.” He was not being dramatic. He was describing what Interpol is actually seeing on the ground.
His full quote:
“Enabled by artificial intelligence, low-cost digital tools and increased global criminal collaboration, we are witnessing the industrialisation of fraud.”
So what does that look like in practice? Fraud-as-a-service platforms on the dark web now sell pre-built fraud kits. Need a convincing payslip from a specific bank? There is a template for that. Need a utility bill from a UK address? Same deal. Template farms on social media sell fraud packages with fake payslips and bank statements for as little as $400. We wrote about this in our post on how social media influencers turned fake documents into a business. It is still happening, and it is getting worse.
Voice and face cloning tools can now replicate a person from mere seconds of genuine audio or video. Interpol documented cases where fraudsters used deepfake audio to mimic corporate executives during live phone calls to authorize wire transfers. And on the low end of the skill spectrum, generative AI has, in Interpol’s words, “demolished barriers to entry.” You no longer need technical ability to pull off sophisticated fraud.
The result is a factory model. Fraud is no longer an artisanal craft. It is produced at scale, distributed through organized networks and sold as a service to anyone willing to pay.
Interpol supported over 1,500 transnational fraud cases since 2024 and recovered $1.1 billion in lost assets. That sounds impressive until you compare it to the $442 billion in losses. That is a recovery rate of about 0.25%.
Digital forgeries now outnumber physical fakes for the first time
This is the data point that should keep compliance teams up at night. According to Entrust’s 2025 Identity Fraud Report, digital document forgeries increased 244% year-over-year. Digital forgeries now account for 57% of all document fraud, surpassing physical counterfeits for the first time ever.
A 1,600% increase since 2021. Four years ago, nearly all fraudulent documents were physical counterfeits (photocopied, scanned, altered by hand). Today, the majority are created entirely on a computer. No printer needed. No scanner. Just software.
Sumsub’s fraud report adds another layer: AI-assisted document forgery using tools like ChatGPT, Grok and Gemini went from essentially 0% to 2% of all detected fakes in 2025. That might sound small, but the trajectory is what matters.
Deepfake fraud surged 1,100% in the US in early 2025. Synthetic identity document fraud rose over 300%.
What does this mean for your business? The old approach of spotting fakes with the human eye is finished. We have been saying this for a while, and the numbers finally back it up.
As Argyle’s research confirmed, 90% of fake documents are now invisible to human reviewers. The fakes are not getting sloppier. They are getting better, faster and cheaper to produce.
Agentic AI: fraud campaigns that run themselves
The Interpol report raises a concept that should worry anyone in compliance or risk management: agentic AI applied to fraud. These are AI systems that can autonomously plan and execute complete fraud campaigns, from reconnaissance to document generation to ransom demands. No human operator needed once the system is running.
This is not theoretical. AU10TIX’s Global Fraud Report declared 2025 the “Year of Machine Deception” and documented agentic AI systems capable of autonomously creating and adapting synthetic identities in real time. Interpol found that AI-enhanced fraud is 4.5 times more profitable than traditional methods.
Dark web marketplaces now offer applications that can clone voices and faces from just a few seconds of real audio or video. Think about what that means for document verification.
A human fraudster might produce a handful of convincing fake invoices per day. An agentic AI system can produce hundreds, each tailored to a specific target, with correct formatting, plausible metadata and internally consistent details. It does not sleep. It does not make careless mistakes.
Synthetic identity fraud alone costs the US an estimated $30 to $35 billion annually. It accounts for up to 80% of new account fraud cases.
Generative AI can now produce a complete synthetic identity profile, documents included, in less than 24 hours. Forget individual fraudsters with Photoshop. This is automated production at scale.
The economics have flipped completely. It used to be expensive and time-consuming to produce high-quality fake documents. Now it is cheap, fast and scalable.
So here is the uncomfortable question: if your document review process still relies on human reviewers checking PDFs manually, how exactly do you plan to keep up?
Why most organizations still have no real defense
Here is the gap the Interpol report exposes without explicitly saying it: most organizations have no automated document verification in their workflow. They check IDs (sometimes). They verify email addresses.
They might run a credit check or screen against a sanctions list. But the actual documents that trigger financial decisions, the invoices, bank statements, payslips and contracts, go through with nothing more than a visual glance.
You would not accept a handshake as proof of identity. So why do so many organizations accept an unverified PDF as proof of payment?
This is exactly what fraudsters exploit. They know that the difference between AI fraud detection and manual checks is the difference between catching a fake in five seconds and never catching it at all.
Consider a typical BEC scenario. A finance team receives an invoice that looks identical to previous invoices from a known vendor. Same logo, same format, same line items. The only difference is the bank account number, buried in the payment details.
A human reviewer who processes dozens of invoices daily will not spot that. An automated system that cross-references document metadata, creation patterns and formatting anomalies will.
This is why we built VerifyPDF. When every invoice, bank statement or credential that crosses your desk could be an AI-generated fake, you need something that checks what humans cannot see: metadata inconsistencies, content layer manipulation, timestamp anomalies and creation tool fingerprints. We process documents in under five seconds and give you a clear risk rating: Trusted, Low risk, Needs attention or High risk. No ambiguity. No “it looks fine to me.”
Interpol’s report makes one thing clear: fraud is not slowing down, and it will not slow down over the next three to five years. Fraudsters’ tools keep getting cheaper. Their output keeps getting better.
The $442 billion figure is not going down. If your organization is still reviewing documents manually, VerifyPDF is where you start. The only question is whether you will catch the next fake before it costs you. Or after.