ComplyAdvantage, the global fintech customer onboarding, transaction monitoring and sanctions screening firm, has published its fifth ‘state of the world’ report on financial firms’ struggles against financial crime.

The report, entitled The State of Financial Crime 2025, is based on a survey of 600 C-suite and senior compliance decision-makers at sizeable financial and fintech firms in Australia, Canada, France, Germany, Holland, Hong Kong, Singapore, the United Kingdom and the United States. Most financiers were in banking – especially private banking – and payments.

Fifty-one per cent of firms surveyed saw compliance costs rise by over 10 per cent in 2022 and 2023 – a trend that seems to be continuing. Organised crime, meanwhile, is playing an increasingly dominant role in global criminality, aided by the continuous easing of trade and travel all over the world, by the rise of the Internet and by communications technology such as smartphones.

The parts of the report that concern us here are those that affect compliance with anti-fraud and anti-money-laundering (AML) laws and rules at private banks and other wealth management firms. The report’s authors, Andrew Davies and Iain Armstrong of ComplyAdvantage, answered questions in a recent webinar that WealthBriefing attended. Among their top concerns was a desire to make the onboarding process as free of inconvenience as possible with the aim of avoiding ‘customer abandonment’ – the bane of compliance in the wealth management sector.

A bull in a china shop 
The report notes that, in recent years, western governments such as the US under President Biden have taken an increasing interest in official corruption in authoritarian or hybrid regimes, especially the misappropriation of public or assets by elite figures within or attached to regimes, often described as ‘kleptocracy.’ The report was published too early to take account of the incoming President Trump’s decision not to use governmental agencies to enforce the US Foreign Corrupt Practices Act 1977, the world’s pioneering piece of anti-corruption legislation.

Armstrong, the consultancy’s regulatory affairs practice lead, told attendees: “Some countries in Europe have only recently brought such laws in. We should expect other radical moves and they will not be uncommon in the years ahead. Nevertheless, when I look across my [compliance] network, the initial reaction to Trump’s FCPA move is: ‘well this has happened, but I'm not going to change my policies.’ When somebody dilutes some regulation it doesn't necessarily mean that you have to change anything in your programmes.”

Delegates at the online conference noted also that Elon Musk had been deregulating and streamlining with extreme prejudice ever since the beginning of the Trump presidency. One remarked: “He could do worse than to harmonise federal and state cyber-regulations, streamline things and cut out the duplication.” This may well happen.

The need to identify the ultimate beneficial owners (UBOs) of bodies corporate is another major job for MLROs at private banks, dealing as they do with UHNW shareholders on a daily basis. In 2024 the US Government enacted updates to the Corporate Transparency Act and the Bank Secrecy Act, particularly regarding beneficial ownership. The report looks forward to greater beneficial ownership transparency in 2025 throughout the world, with an international drive underway to improve the accuracy of UBO information contained in corporate registries. This, however, is now out of date in respect of one country. On 2 March Trump suspended all enforcement of the Corporate Transparency Act against US citizens and domestic reporting companies.

Meanwhile, on 31 January, Trump signed an executive order to ensure that whenever an agency promulgates a new rule, regulation or guidance, it must earmark at least 10 existing rules, regulations, or guidance documents to be repealed. This is a turbo-charged version of the ‘2-for-1’ order of his first presidential term.

Michael Knight-Robson, a director at the accountancy firm of BDO, went as far as describing President Trump as “a bull in a china shop” and warning that “some jurisdictions like the USA might pull away from some of the Financial Action Task Force’s 40 recommendations.” These are the global anti-money-laundering (AML) standards which have been in place, in some form or another, since 1989.

Factors that hamper compliance
ComplyAdvantage asked the respondents a simple question which elicited many answers from each person: What are the main limitations to your organisation's current approach to financial crime detection? The results were as follows.

Siloed datasets....................................………..45%
Lack of real-time visibility into risks..................45%
Data quality.......................................……….…44%
Sanctions/watchlist screening..........................43%
Ongoing monitoring.................................……..41%
Lack of rule flexibility...........................……......41%
Unclear regulatory expectations.......................35%
Lack of effective case management...................5%
There are no current limitations....................…..1%

It comes as no surprise that 99 per cent of the survey’s participants acknowledged the existence of barriers to successful crime detection. As many were from multinationals, including private banks, it was also inevitable that data silos – collections of data held by one group that are not easily accessible by others in the same organisation – were a major frustration for compliance officers and money-laundering reporting officers (MLROs). Forty-five per cent of respondents said that they were sometimes unable to draw links between pieces of related data. ComplyAdvantage put this down to an inability to collect data of high quality and make inferences from it quickly. One ComplyAdvantage AML expert said that this was a problem at many of the large banks with which he conversed.

Artificial intelligence
For the second year in a row, the survey showed that the participants were fairly comfortable with their use of AI, which one delegate defined as “the use of technology to quantify large datasets.”

Intriguingly – especially in light of respondents’ stated understanding of AI regulations and the importance of transparency – 46 per cent that said they were very comfortable and 45 per cent said that they were somewhat comfortable with “compromising explainability in return for increased automation and efficiency.”

This is not what regulators want to hear. ‘Transparency and explainability’ occur when the developers (software vendors) and deployers (private banks, asset managers, networks etc) of AI understand the ways in which their systems work and are able to explain them – clearly, easily and credibly – to regulators, customers and anyone else who might be affected by those systems’ decisions – especially adverse ones.

In March 2023, HM Government published its AI Regulation White Paper which set out five principles for regulators to interpret and apply within their remit. One of these was "appropriate transparency and explainability." Moreover, the paper said that this rule underpinned all the others. The report therefore implies that any financial firm that is not able to explain the workings of its AI programs to regulators is playing with fire.

AI vendors are often not very good at explaining their own products to compliance officers. A successful sale depends to a great extent on the purchasing bank (or other wealth management firm) asking the vendor a blizzard of questions about the applicability of the software to its particular data sets and corporate habits. In too many instances, this does not happen. Also, not every AI vendor lets the bank’s compliance department test its software before the purchase.

With KYC, firms now have AI-powered entity resolution, a process that identifies and matches records across multiple data sources to create a golden record of key business entities, to help them verify people's identities. Knight-Robson explained: “It’s not just your typical driver's licence, passport, scanned copy, or using Experian or Equifax...it’s the use of multiple sources to help with that identification.”

AI is also used in trade financial AML because it can analyse documents faster than humans. Various delegates at financial firms commented that they were thinking of using it to “empower our analysts” as well. Every SAR has a space in which the MLRO can write a narrative that explains why the activity is suspicious and firms are now using AI to write these as well. Davies thought that the cost of using such techniques was going down.

Can AI ever be trusted to take AML decisions on its own? Armstrong thought that it might be possible.

“You need to think about where you have high-volume, low-risk scenarios that lend themselves to that rapid yes-no decision making. Clearly there are more complex scenarios where AI should only really be supporting human decision-making. We see that a lot with name screening for adverse media and political exposure, where context is everything. We've noticed that you can get an increase in speed or the time taken to work an alert if you have the right level of context for an AI-generated alert.”

Davies added that the use of AI was not mandated in any jurisdictions of which he was aware.

Delegates were reminded that their criminal adversaries were using AI as well. They, of course, are not limited by considerations of data privacy, anti-corruption legislation or agencies such as the US Consumer Financial Protection Bureau, which President Trump is trying to close down.

Countries on the FATF grey list 
At the end of last month the FATF updated its so-called grey list of "jurisdictions under increased monitoring" by adding Kenya and Namibia and removing Barbados, Gibraltar, Uganda and the United Arab Emirates.

The survey asked each company to pick the country on the list that concerned it the most. The results, listed from the country that inspired the most worry to the one that inspired the least, are as follows: The Philippines; South Africa; Kenya; Nigeria; The Democratic Republic of Congo; Burkina Faso; Syria; Cameroun; Vietnam; Mozambique; Monaco; Bulgaria; South Sudan; Venezuela; Croatia; Haiti; Namibia; Senegal; Mali; Tanzania; and Yemen.

Here and there
The conference noted that the European Commission (the nearest thing that the European Union has to an executive branch) has, according to POLITICO, decided to withdraw its proposal for a Financial Data Access (FIDA) regulation as a result of opposition from the financial services sector and the French government. This was hailed as a missed opportunity for some useful 'open finance' rules that would have facilitated the sharing of financial data between banks and HNWs' third-party service providers through Application Programming Interfaces. One delegate said that, if true, it would be "a blow against modernisation in the EU."

The conference heard that in Australia on 31 March, changes to the 'tipping off' offence will come into effect. This offence is committed when a bank discloses information to a customer that is likely to prejudice an investigation.

In May, dealers in high-value goods and services will begin to report sanctions breaches to HM Revenue and Customs in the UK. The UK is a major international hub for the trade of high value goods, including art, antiques, luxury cars, precious metals and gemstones, and for investment in wines and whiskies to name a few. In 2023 it accounted for 17 per cent of the world market in art sales, according to HMRC.  

On 1 September, a new fraud prevention offence will take effect in the UK. A private bank or other wealth-management firm may be held criminally liable whenever its employee, agent, subsidiary or other “associated person” commits a fraud with the intention of benefiting the organisation.

Examples may include dishonest sales practices, hiding important information from clients or investors or dishonest practices in financial markets. In the event of prosecution, the bank will have to prove that it had reasonable fraud prevention measures in place at the time.