8 red flags that signal document fraud—and why AI-powered forensics catches what your eyes miss
78% of landlords verify income documents by eye alone. Here's what they miss:
Result: Average fraud loss: £23,500 per incident (eviction costs + lost rent + court fees).
Real payslips vary. They have strange deductions, varying tax codes, and odd net amounts. If all months show exactly £2,000.00 net, that's fabricated.
Why it works for fraudsters: When creating fake payslips, people guess round numbers. Real employers generate complex calculations.
Payslips have consistent fonts. If some text is Arial and other bits are Helvetica—or if numbers look pixelated while text is clean—sections have been edited in image editing software.
What to look for: Different font weights, blurry numbers, misaligned text, or characters that don't match employer letterhead.
PDFs store creation dates. If a "recent" payslip shows a creation date from 5 years ago, or a modification date 1 hour before submission—something's wrong.
How fraudsters miss this: Most don't even know PDF metadata exists. AI tools always check it.
If January payslip shows £2,500, February shows £3,200, and March shows £2,650—ask why. Real salaries are stable unless it's a new job.
Why this matters: Fraudsters create documents independently. They forget to keep salary amounts consistent across months/years.
Payslips list specific employer details: company registration, PAYE reference, address. If it's vague ("ABC Ltd", "Company House", generic addresses), it's likely fake.
Red flag: Can't verify the employer? It probably doesn't exist.
UK tax codes follow strict patterns (e.g., 1257L for 2024/25). If the tax code doesn't match the employee's circumstances—or is entirely generic—it's fake.
What to check: Does the tax code make sense for the salary level? Are deductions typical (PAYE, NI, pension)?
If a candidate worked at "Company A" and "Company B" but both payslips have identical letterheads, fonts, and layouts—they're from the same source file.
Red flag: Copy-pasting from the same template.
Every payslip has a unique reference number (employee ID, payslip number). If the pattern is sequential or suspiciously uniform across years, it's likely fabricated.
What real patterns look like: Random alphanumeric codes, company-specific formats, irregular sequences.
Honestly? No. Flags #3 (metadata) and parts of #1, #2, #6 require technical analysis tools. A few hours in Photoshop and most landlords can't tell.
The gap: 80% of modern fraud is invisible to manual review. AI tools scan thousands of data points simultaneously.
Forensic AI systems check:
This doesn't replace human judgment, but it eliminates 90% of common fraud in seconds.
The Scenario: Tenant claims £3,500/month salary, affordability: ✓
Red Flags Found by AI:
Verdict: HIGH RISK—Document tampering detected. Recommend: Employer verification call, bank statement cross-check, or reject application.
Checking payslips manually takes 20-30 minutes and catches maybe 40% of fraud. AI-powered forensics:
At , it costs less than 1 hour of lost rent.
Run a Forensic ScanRead more about tenant verification in our blog, or use our forensic scanner to check documents in real-time.
← Back to Blog