Money laundering in the UK was estimated at £154.83bn in 2025, equivalent to 5.16% of GDP, in a new international index developed by financial crime technology company Napier AI with research partner GlobalData. The sterling figure converts the report’s US-dollar estimate at a rate dated 21 October 2026. It represents modelling of illicit activity, not transactions proved to involve money laundering.
Napier AI says the proportion of UK GDP represented by estimated money laundering declined from 5.53% in 2024 to 5.16% in 2025. Britain improved ten places to rank 24th among the 42 markets assessed. That improvement in the model’s comparative ranking does not establish that UK financial institutions have eliminated the underlying risks.
The company’s AI / AML Index 2026–2027 estimates the annual cost of UK anti-money laundering compliance at £5.88bn, down 7.7% from £6.37bn. It identifies a further £3.08bn in potential savings associated with applying AI to relevant compliance processes. Those projected savings depend on the report’s methodology and assumptions and should not be presented as expenditure that banks can immediately remove from their budgets.
The report’s global analysis, spanning 42 jurisdictions, compares estimated criminal financial activity, compliance costs, effectiveness and the regulatory environment. The results differ substantially across markets. Comparisons need care because detection quality, reporting requirements, financial-sector size and the visibility of illicit transactions vary between countries.
Greg Watson, chief executive of Napier AI, argued that sanctions changes and cross-border criminal activity continue to create opportunities for offenders to exploit weaknesses in financial systems. The company identifies activity linked to money mules, human exploitation, trafficking and high-value goods among the typologies informing its assessment of UK exposure.
For banks and payment companies, anti-money laundering programmes involve customer due diligence, transaction monitoring, sanctions screening, investigations and reporting suspicious activity. These functions are costly partly because alerts may require analysts to examine multiple datasets and determine whether apparently unusual transactions have legitimate explanations. A high volume of false positives can absorb time without proportionately increasing the detection of real misconduct.
AI can potentially help by identifying patterns that traditional rules miss, prioritising cases or summarising data for an investigator. Its value depends on data quality, explainability, privacy and the ability to reconstruct why a particular decision was made. An automated system that produces an opaque risk score can create new governance difficulties, especially when institutions must justify actions to regulators or customers.
The index found that regulatory requirements remained the principal investment priority for 64% of surveyed institutions, compared with 20% prioritising new technology and 12% hiring or training specialists. That distribution suggests many financial firms continue to evaluate technology investments primarily through their compliance obligations rather than treating AI as a substitute for them.
Napier AI also reported that 68% of financial crime professionals believed their existing anti-money laundering engines could support AI’s data requirements. Such self-assessments indicate perceived readiness, not independently tested compatibility or proof that institutions have access to every dataset needed for effective detection.
Janet Bastiman, Napier AI’s chief data scientist, emphasised the importance of auditable automated decisions and human judgement. That approach reflects the tension between reducing investigation workloads and preserving accountability for decisions that may restrict customer access to financial services or involve sensitive personal information.
A further complication in interpreting international comparisons is that the underlying criminal market is deliberately hidden. Changes in reported activity may reflect better detection, different modelling inputs or real changes in illicit flows. The index offers a structured way to compare estimates, but it is not a direct audit of money entering the financial system. Institutions should assess operational controls using their own validated detection and case-outcome measures.
The index provides an indication of the scale that its authors attribute to money laundering and the potential value of better analytical tools. It does not settle the true volume of hidden criminal activity. For regulated organisations, the more practical questions are how models are validated, how errors are challenged and whether deployment produces measurable improvements in detection without weakening customer protection.
You must be logged in to post a comment.