Swapped bank details
IBANs are checksum-validated and compared against the bank details a supplier has used before. Invoices that share a bank account with other vendors are linked.
Invoice fraud detection before you pay.
Docklands AI is invoice fraud detection for accounts payable teams. It checks every supplier invoice for changed bank details, duplicates, fake vendors and AI-generated documents before payment.
app.docklands.ai/invoices/INV-20418
KESTREL SUPPLY CO.Invoice INV-20418 · Net 30
Hydraulic hose x121,140.00
Fittings kit386.50
Delivery45.00
TOTAL1,932.80
Pay to IBANIE29 ••• 4471Shared with 2 vendors
- ⋔IBAN shared with 2 vendorsnetwork
Invoice checks 214 run
- VAT number formatpass
- Line items add uppass
- IBAN checksumpass
- Business at addressfail
- Invoice number sequencepass
- VAT rate matches categorypass
Every invoice before it is paid.
Supplier invoices
PDF or photo, one invoice or a batch. Every number, date, bank detail and address.
Credit notes and re-issued invoices
Checked like the originals, so a changed total or bank account on a re-issue stands out.
Receipts and proof of payment
Anchor documents that show what was actually paid.
Supporting attachments
Quotes, delivery notes and timesheets attached to the invoice.
What invoice fraud detection has to catch.
Once an invoice is paid the money is usually gone. These are the patterns we see most.
Duplicate and re-submitted invoices
The same invoice sent twice, or edited and sent again, is matched across your history and flagged. Invoice number sequences are checked too.
Vendors with no trading address
Addresses, VAT numbers and phone numbers are checked, so a supplier that does not trade anywhere is flagged.
AI-generated invoices
Invoices made with image generators look real to a person. Our own models detect the fingerprint they leave.
Line items priced above market
Every line is priced against the market, so overcharging stands out before approval.
Totals that do not add up
Line items, VAT and totals are recalculated, and the VAT rate is checked against the category.
From one document to a decision. In seconds, not days.
Every file goes through the same four stages. Your team only sees what needs a human.
Ingest
API, email forwarding or upload. Receipts, invoices, PDFs, phone photos: one file or a hundred.
Inspect
Custom models for AI fakes, tampering and edits, plus 200+ checks on every invoice.
Cross-reference
A holistic review compares every file on the claim and clears flags the full picture explains.
Decide
A risk score with evidence pinned to each document, sent to your system or inbox.
Why metadata checks miss invoice fraud.
- An invoice PDF can be rebuilt from scratch with clean metadata. The fraud is in what it says.
- Docklands AI inspects the invoice image for AI generation, manual tampering and digital edits.
- 200+ checks run on every invoice: line items add up, VAT rate matches category, invoice number sequence, IBAN checksum and more.
- Results arrive with the evidence located on the document, ready for the approver.
What the metadata says
- DeviceiPhone camera
- Created14/09/2026 18:02
- Edited withnone
- VerdictLooks clean
What the pixels say
- Generator fingerprint0.97
- Total re-written0.91
- Cloned digits2 regions
- VerdictHigh risk
See what Docklands AI finds in your own documents.
Your first 1,000 documents are checked free.
Vendor fraud detection
Vendor fraud takes many forms: a supplier that does not exist, a real supplier whose bank details have been swapped, or a group of vendors that are really one business billing several times over. What they have in common is that the invoice looks ordinary to the person approving it.
Docklands AI checks whether each vendor trades where it says it does, validates its VAT number and bank details, and links invoices that share a bank account, phone number or vendor with other suppliers. Several "different" vendors paid into the same account show up as one network, with the evidence attached.
Accounts payable fraud prevention without slowing payment runs
Accounts payable teams cannot inspect every invoice by hand, and fraudsters know it. Docklands AI checks every invoice automatically, with a median verdict time of 43 seconds, so approvers only stop on the ones that need a person.
Invoices arrive the way they already do: by email forwarding, REST API, SharePoint or Google Drive. There is no new upload portal for your team, and when an invoice is flagged the report lands in your inbox with the evidence attached.
Security
Built for claims data.
Your documents stay protected, auditable and where you need them.
Multi-factor login
MFA with backup codes, trusted devices and guided recovery.
US or EU data residency
Choose where your documents are processed and stored.
Full encryption
Every document and every finding is encrypted.
Custom data retention
Keep documents as long as your policy requires, and no longer.
SOC 2 Type I
Independently audited security controls.
ISO 27001
Certified information security management.
FAQ
Invoice fraud detection questions.
What is invoice fraud in accounts payable?
Invoice fraud in accounts payable includes AI-generated invoices, swapped bank details, duplicate invoices sent multiple times, inflated line items priced above market rates and fake vendors with no real business address. These schemes cost organizations millions in undetected fraudulent payments.
How can you tell if an invoice is fake?
Fake invoices often show AI generation patterns in grain and light, digital edits visible at pixel level, mismatched bank details, vendors with no trading address or inflated prices above market rates. Advanced fraud detection software runs 200+ automated checks on each invoice.
What does Docklands AI check on a supplier invoice?
More than 200 checks covering numbers, dates, bank details, VAT and addresses, plus image analysis for AI generation, tampering and digital edits.
Can it catch changed bank details?
Yes. IBANs are checksum-validated, and invoices that share a bank account with other vendors or claims are linked and flagged.
Does it detect duplicate invoices?
Yes. Duplicate and re-submitted invoices are flagged, including ones that were edited before being sent again.
How do invoices get to Docklands AI?
By REST API, email forwarding, SharePoint or Google Drive. There is no new upload portal for your team.
How long does integration take?
Under 30 minutes via REST API, or send documents by email with no integration at all.
Still have a question? Ask us on a call. Book a call
Send us the invoice you don't trust.
We'll show you what's been faked, what's been changed and what's been overcharged, before it's paid.