Compliance-first AI: Why you can’t just spend your way to smarter AML

Greg Watson, the CEO of Napier AI, explains why compliance-first AI, not more spending, will define the next phase of AML innovation.

For years, financial institutions have tried to outspend financial crime. Yet despite an estimated USD 274 billion (USD) spent globally each year on compliance, money laundering losses have surged to more than USD 5.5 trillion (USD) – equivalent to nearly 5% of global GDP, according to the latest Napier AI / AML Index 2025-2026.

The problem is not a lack of investment. It’s a lack of intelligence in how that investment is applied. 

Now, as financial institutions experiment with generative and agentic AI, it is important to keep in mind that, when innovating, compliance comes first. The purpose of AML automation in anti-money laundering (AML) is innately human: to stop the flow of illicit money that funds criminal activity. It’s not about teaching machines to think like humans, but about building systems that help humans make better compliance decisions – faster, more accurately, and with greater explainability.

The Napier AI / AML Index reveals that leading markets like Spain, Canada, and the UAE are proving this point. By adopting explainable, compliance-first AI, these nations are spending more efficiently while achieving stronger AML outcomes. In these markets, the ideal ratio of AML spend to money laundered now sits between 1.36% and 3.36% - a clear sign that well-implemented AI doesn’t just cut costs; it raises capability.

This graph shows the ratio as a percentage for each country of the cost of compliance to money laundered. The ‘ideal’ range of compliance spend to money laundered is assumed to fall between 1.36 and 3.36%, based on a weighted average of the compliance spending for the higher-performing markets in terms of ML losses and effectiveness of AML. 
Source: Napier

The rise of AI-assisted analysts

AI is already speeding up routine processes once considered immovable. Compliance teams are testing models that scan legacy documentation, extract relevant rules, and answer analysts’ questions in seconds.

This shift isn’t theoretical; it’s already happening. In recent global research conducted for the Napier AI / AML Index 2025-2026, 73% of financial institutions described AI as ‘very useful’ for post-investigation model learning and pattern analysis, while over two-thirds (69%) said AI is already highly effective in transaction flagging.

Used wisely, these tools are shortening investigative cycles that once took days. Used recklessly, they risk leaking sensitive customer data or propagating bias.

The AI trust test

The key difference lies in explainability. Every compliance decision must be traceable: how was a rule interpreted, what data was used, and why was a conclusion reached? Firms want explainable AI in finance. It should not only be powerful, but auditable – capable of showing its work to both global AML regulation and internal reviewers.

This is the AI trust test. And across the globe, work is being done to innovate with AI, with transparency needed to really understand its output.

• UK AML regulation: In the UK, regulators have brought their markets out of the experiment phase and into tangible AI innovation, launching the FCA’s Live AI Testing and the Supercharged Sandbox to help firms demonstrate transparency and oversight while trialling AI systems in production environments.
• US AML regulation: In the US, a global leader in AI for AML, regulators continue to refine frameworks to ensure that AI-enhanced AML models can be audited – balancing innovation with the demand for evidential accountability.
• France AML regulation: In France, the supervisory regime is technically advanced but cautious in rolling out AI, reflecting ‘regulatory conservatism and supervisory caution’, with guidance emerging for large language model governance in financial compliance contexts.

Keeping up with the regulators

Despite this progress, many firms feel they’re still behind the regulator. There’s a strong appetite for standards that define what ‘good’ AI looks like in financial crime compliance –  covering parameters, evidential requirements, and the depth of explainability expected. As regulators themselves start experimenting with agentic AI in AML, industry collaboration will be essential to ensure safe innovation.

Until then, the most successful institutions are adopting a compliance-first AI approach: using AI where it adds demonstrable value, keeping humans in the loop, and documenting every decision to ensure transparency and trust. 

Practically, this means financial institutions must build AI models that ensure every output can be traced, tested, and defended to regulators and boards alike. The firms that can explain their AI will be the ones trusted to use it.

Global milestones: Regulation, deregulation, and innovation in AI. Regulation is no longer moving in one direction. Some markets are tightening, expanding AML and AI oversight to close long-standing gaps. Others, like the US, are pulling back in certain areas under a deregulatory agenda, even as they tighten rules elsewhere. 
Source: Napier

From compliance burden to strategic value

The ultimate goal isn’t to replace human compliance expertise but to amplify it. Napier AI’s AML Index 2025-2026 revealed that AI adoption could save regulated firms USD 183 billion globally each year. This is by reducing false positives, accelerating investigations, and surfacing insights hidden in data – but only when built on a foundation of governance and accountability.

Firms that get this right will turn compliance from a regulatory necessity into a strategic advantage for faster onboarding, sharper risk detection, and higher customer confidence.

Compliance-first AI isn’t about spending more; it’s about thinking smarter. Because in the fight against financial crime, the winners won’t be those who spend the most – but those who understand the problem best.

See how AI is reshaping AML effectiveness across 40 global markets. Read the Napier AI / AML Index 2025-2026

About author

Napier's AI CEO Greg Watson brings knowledge and experience gained over the past 25 years in the financial services industry, including roles at HSBC, where he built a client-facing organisation to manage client onboarding and due diligence, spanning the entire product set of GBM across 35+ countries. Greg’s specialty is the challenges facing global financial institutions in client onboarding and lifecycle management, and how financial institutions can build more effective digital processes.  A frequent contributor to forums and publications, Greg is a staunch advocate of leveraging modern technology to aid in the fight against financial crime while making the customer experience as frictionless as possible.

 

About Napier AI

Napier AI is a RegTech delivering anti-money laundering and financial crime compliance software to financial institutions, payments, and wealth & asset management firms. Napier AI designs and engineers technological innovation to make a measurable difference in driving down financial crime. Trusted by over 100 institutions worldwide, the company’s platform, Napier AI Continuum, is transforming compliance from a legal obligation to a competitive edge.