Introduction

Artificial intelligence-backed technology is making waves in adverse media screening, furthered by advancements in general machine learning, large language models and natural language processing. With capabilities that offer operational efficiencies and improve anti-money laundering efforts, we examine how the availability of new technologies impacts organisations practically, assists compliance teams with their biggest challenges, and paves the way for the future of adverse media screening.

Ripjar has already examined in-depth some of the common challenges for senior compliance staff in our white paper, the Chief Compliance Officer’s Playbook, where technology is identified as a strategic tool for forward-thinking compliance leaders. Tools that process a wide range of data smartly and efficiently are no longer seen as ‘nice to have’ but ‘must have’ investments, allowing them to harness the strengths of human and machine capabilities. But how and to what degree do firms use advanced technology in adverse media screening? 

We’ve surveyed 205 compliance professionals in firms across Belgium, the Netherlands, Luxembourg, Sweden, Finland, Germany, France, Italy, the United Arab Emirates (UAE), and the United Kingdom. Showcasing the diverse use cases of adverse media screening, which is certainly not only limited to financial institutions, these companies operate in diverse industries, including government, services, communications, construction and manufacturing, software and technology, real estate, retail and wholesale trade.

We surveyed  respondents about their use of adverse media screening, including common challenges, how they use technology, and how automation helps with risk detection.  Across the themes identified, we found that firms believe the common challenges they face, which encompass regulatory changes, the complexity of using new technology solutions, and a potential increase in false alerts — are poised to get harder.

Overwhelmingly, respondents believe that automation and technology solutions deliver higher confidence in risk detection, emphasising the importance of pushing on through perceived challenges with technology implementation to achieve better results and support a risk-based approach. 62% of respondents believe they rely too heavily on manual processes to screen for high-risk individuals, highlighting that technology adoption holds much-untapped potential and will likely continue to be a focus area for investment.

This report draws out four key themes from our research, contextualising findings in the anti-financial crime regulatory landscape. By examining responses, we look at the opportunities and challenges the future of artificial intelligence-backed technologies holds for the future of firms’ anti-financial crime strategies. 

---

Common Challenges

In a fast-moving regulatory environment, there is a wide range of regulatory and logistical challenges for organisations to tackle when undertaking adverse media screening. Firms we surveyed reported key challenges ranging from technology adoption and resource constraints, to streamlining screening processes and staying on top of regulatory changes.

Overall, many firms predict the current challenges they face, such as resource and budget constraints, false positive levels and technology and data integration, will continue to grow. 47% of firms believe that the challenge of keeping in step with regulatory changes will increase. Specifically, when broken down by industry, this is the leading concern for those operating in financial services.

As anti-money laundering and counter-terrorism financing regulations and enforcement become more stringent, firms must explore how their controls and processes become more efficient and effective. In the United Kingdom, the Financial Conduct Authority (FCA) more than doubled the fines issued, from 10 in 2021 to 26 in 2022. In the European Union, the implementation of the most recent Sixth Anti-Money Laundering Directive (6AMLD) significantly expanded the list of money laundering predicate offences, including cybercrime, environmental crime, migrant smuggling and sexual exploitation. This means that financial institutions must now account for these predicate offences in their open-source search processes.

The challenges felt by each firm were interlinked to the maturity of their current technology adoption. Overall 41% of respondents thought that the complexity of implementing new technology solutions will remain one of their top challenges for the next 12-24 months. Unsurprisingly, for those firms that use only manual processes for their adverse media screening, the complexity of implementing new technology solutions was identified as the top challenge. These findings suggest that firms which have not yet adopted technology for their adverse media screening may be considering doing so, yet have held back by perceived fears of difficulty.

Looking forward, resourcing and lack of workforce are front and centre for many. 42% of firms believe resourcing will still be challenging over the next 12-24 months. Deploying technology reduces time-consuming manual tasks, streamlines processes, and creates efficiencies in a firm’s adverse media approach. As a result, resource challenges can be alleviated while ensuring efforts are directed to higher-risk activity. 40% of firms believe that the challenge of irrelevant alerts will increase, highlighting the need for technology solutions that employ measures to reduce false positives. Sophisticated tools can make sense of unstructured data, such as using secondary identifiers for identity resolution, which can significantly improve the false alert rate and alleviate the burden for firms. 

For firms in the middle of their tech journey – meaning those that use primarily manual processes with some technology – the challenge of receiving too many alerts was forecasted to increase, suggesting that they have not yet fine-tuned their methods to reduce irrelevant alerts compared to other companies that have implemented more technology.

Deploying technology reduces time-consuming manual tasks, streamlines processes, and creates efficiencies in a firm’s adverse media approach.

---

The use of technology in adverse media screening processes

With vast differences in adverse media screening methods used by organisations, technology flexibility and integration was once again a hot topic for survey respondents. To better understand the impact of technology solutions, we also explored how firms’ confidence in their screening results varied between methods used.

For those using technology in adverse media screening, current approaches include using a combination of in-house and vendor solutions (34%), an internal solution designed in-house (31%), or an external solution from a vendor or supplier (28%). These figures highlight the preference of firms to adopt a custom approach to technology adoption. There is no one-size-fits-all, and each company has different requirements contingent on its size, jurisdiction, or type of business. Having technology solutions that can work with other solutions and are easy to integrate is paramount.

Where technology is deployed, 71% of respondents said they use artificial intelligence and machine learning as a core component of their adverse media screening. Interestingly, there are some regional nuances, with the UAE, Benelux, and Nordic countries using this technology more. In many of these countries, there is growing acceptance of sophisticated technologies, exemplified by the recent victory of Dutch challenger bank Bunq, which won a case against the Dutch Central Bank over artificial intelligence use for anti-money laundering.

Staggeringly, 20% of respondents are still conducting adverse media screening via entirely manual processes. For firms that use at least some manual screening, search engines are the most commonly used manual process in adverse media screening (49%), followed by manual searches of a firm’s own database (41%).

This is interesting, given the Wolfsberg Group’s guidance from 2022 that states that using a search engine like Google, or Boolean searches, is no longer sufficient given the quantities of data to sift through and the potential for human error. The paper also warns that search engines are algorithm-driven, meaning they may inadvertently de-prioritise relevant information and are not “designed to assess financial crime risk.”

In terms of confidence, respondents that use technology-led solutions for all their adverse media screening are most confident (100%) in its results. Those that mostly used technology-based processes with some manual were 82% confident, whereas those that used mainly manual or all manual processes had confidence levels of 73% and 68%, respectively. This data shows that the more technology-led processes a firm uses, the higher the confidence in its ability to screen for negative media.  

This confidence makes sense given that the use of technology has been explicitly recognised by the global money-laundering and terrorism-financing watchdog, the Financial Action Task Force (FATF). In a report, the FATF discusses the potential of technological innovation, with artificial intelligence and machine learning tools specifically allowing firms “to carry out traditional functions with greater speed, accuracy, and efficiency”.

---

Moving away from manual processes

Despite regulators and watchdogs promoting the benefits of adverse media screening technology, our survey highlighted a surprising, industry-wide over-reliance on manual screening processes. With a majority of firms reporting a desire to increase their use of screening automation, we explored the perceived challenges and advantages of moving away from manual screening processes.

Manual processes are time-consuming, error-prone, and have limited scalability for firms. The Monetary Authority of Singapore (MAS) specifically called out the shortfalls of manual screening, especially when large amounts of information are involved, noting the process is prone to human error and delayed performance. Comparatively, technology solutions for adverse media screening are faster, more scalable, and more consistent.

Despite their inefficiencies, organisations still largely rely on manual processes, with 67% of firms being somewhat reliant and 17% significantly reliant. Only 17% of respondents are not reliant on manual processes at all. For those that rely on manual processes either somewhat or significantly, 65% want to increase their automation solution capabilities while still retaining some manual processes. A notable 25% of respondents said they would prefer to be completely automation-solution based. These results suggest that firms have positive attitudes towards technology and wish to implement more technology-based solutions for their adverse media screening. 

To further support this, 62% of respondents that complete at least some manual adverse media screening processes believe they rely too heavily on manual processes to screen for high-risk individuals.

When it comes to adopting new technology to improve screening for high-risk individuals and move away from the inefficiencies of manual efforts, the top challenges firms face are centred around budgetary challenges, vendor selection, technology integration, and internal adoption of solutions.

These challenges highlight the importance of having a clear technology roadmap and selecting the right vendor to support integration.

---

Automation and technology for higher confidence in risk detection

Regulators worldwide require firms to take a risk-based approach to screening, which makes adverse media screening a vital tool for gaining a more holistic view of customer risk. With the increasing buzz around large language models such as ChatGPT, we sought to understand survey respondents’ confidence in these new technologies to support screening.

Screening for high-risk individuals goes beyond simple name searching against sanctions lists. Increasingly, and as evidenced by the relatively recent onslaught of Russian sanctions, screening for risk often includes untangling complex relationships to understand corporate and interpersonal relationships. Adverse media screening uses ever-growing amounts of online data and is essential to a risk-based approach.

The FATF advocates for a risk-based approach, codifying it in the 2000s. The risk-based approach allows firms to have a degree of flexibility in managing their risks. This model tailors the compliance programme based on each individual customer’s risk level, allowing for better allocation of finite resources on higher-risk customers. However, because this model centres on the individual customer’s risk profile, it makes knowing the customer throughout the entirety of a relationship even more vital. Adverse media screening is an essential part of a risk-based approach because it allows financial institutions to leverage a wide range of sources of information to gain a more holistic insight into potential financial crime risks throughout the relationship.

According to our research, technology-based processes improved confidence in identifying risk across all categories surveyed. Manual-led firms were under average confidence in most categories.

The risk categories as part of the research include: 

• Participation in an organised criminal group and racketeering
• Terrorism 
• Trafficking in human beings and migrant smuggling
• Sexual exploitation
• Illicit trafficking in narcotic drugs and psychotropic substances
• Illicit arms trafficking
• Illicit trafficking in stolen and other goods
• Corruption
• Fraud 
• Counterfeiting of currency
• Counterfeiting and pirating of products
• Environmental crime
• Murder and grievous bodily injury
• Kidnapping , illegal restraint, and hostage-taking
• Robbery or theft
• Smuggling
• Tax crimes relating to direct or indirect taxes
• Extortion
• Forgery
• Piracy 
• Insider training and market manipulation
• Cybercrime

Looking ahead at future technology opportunities, most firms have a positive view of using large language models, such as ChatGPT, as an additional capability for adverse media screening. 50% of respondents are potentially considering exploring the use of these models in the future, 34% are already exploring these models, and 2% are already using these models. Only 14% of firms are not considering using large language models to support adverse media screening.

Regarding confidence in large language models, 58% of respondents are somewhat confident in them, 23% are not very confident, and 12% are very confident. Given that ChatGPT is a new tool, having only been released in November 2022, trust in its abilities will likely increase as the technology improves and another iteration is released. Technology vendors may eventually incorporate elements of this technology into their adverse media screening tools. However, despite their advanced capabilities, large language models are presently insufficient for a firm’s screening purposes, as they can give outdated and fabricated information and potentially introduce bias into the process.

In addition to their potential for adverse media screening, language learning models can also be used by criminals to advance fraud and scale their financial crime efforts. As explored in a report by Europol, the availability of large language models has the potential to increase social harm in many areas, and will be a point of interest for law enforcement. This reinforces the need for firms to fight innovative technology-led bad actors with innovative technology-led solutions.

---

Conclusion

Challenges with adverse media screening greatly depend on where firms are in their technology-adoption journey. Yet, the overarching challenge, perceived to grow in the next 12-24 months, is keeping up with regulatory changes. Despite differing levels of reliance on manual processes, only 6% of respondents said they did not use any adverse media screening technology, with the majority of firms wanting to increase their automation solution capabilities in some capacity. These findings suggest that technology supporting adverse media screening is the way forward, and firms will continue adopting it to improve risk management. In a challenging regulatory environment, embracing technology offers distinct advantages to optimise compliance.

Strikingly, technology-based processes improved confidence in identifying risk — with a staggering 100% of firms using technology-led solutions for their adverse media screening being most confident in its results. Looking forward, the advent of mainstream large language models like ChatGPT will likely bring more opportunity, trust, and overall interest in using technology for negative news screening. However, because large language models have weaknesses, such as sharing fake, outdated or biassed information, they cannot serve as an alternative to screening, likely only becoming a part of a technology vendor’s more extensive toolkit in adverse media screening solutions.

Firms that have not yet implemented sophisticated adverse media screening technology will benefit greatly from doing so, as industries shift away from manual processes and towards efficiencies afforded by artificial intelligence and machine learning. These include the ability to make sense of vast amounts of unstructured data, use natural language processing to identify risk, analyse information in multiple languages, and prioritise alerts for analysts. Harnessing these technologies empowers firms to navigate the evolving landscape with agility, and achieve proactive risk management with efficiency.