Eu FLag
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No101022004

TRACE final conference in Vienna: achievements and challenges in the AI, law enforcement, and anti-money laundering fields

The TRACE final conference marks the end of a three-year research project, dedicated to creating an AI tool to support law enforcement and financial intelligence units engaged in the fight against money laundering. Partners and guest speakers from across the worlds of research, policing, policy and security gathered in Vienna for two days of debate on the opportunities of technology and international cooperation for seizing illegal assets and prosecuting offenders. The diversity of guests and perspectives gathered for the event was striking, and their input was invaluable.  

The first keynote came courtesy of Anthony Cook, from the USA’s Internal Revenue Service Criminal Investigation division and Andrew Lonas from the US Embassy in London. They stated that all complex financial crime is international nowadays, thus formal and informal cooperation across borders is imperative. As experienced ‘tax police’, Anthony and Andrew discussed definitions of tax havens that involve jurisdictional vulnerabilities such as weak enforcement, inadequate laws, or weak mechanisms for asset recovery; dimensions which contribute to the prospect of a territory qualifying as a tax haven. Resources are limited, and organisations like the IRS are seeking to leverage public awareness as well as technology to try and boost asset recovery. An escalation in seizures even from 1% to 2% of these illicit funds would be revolutionary, amidst challenges such as the increased centrality of cryptocurrencies. Citing the first batch of TRACE policy briefs, Anthony noted that the IRS has made significant steps with international partnerships and the use of new technologies (including AI and Blockchain tools) to evaluate evidence and assess suspicious transaction reports and geographic risk. A new wave of IRS funding from the American Inflation Reduction Act, which was revised to allot some 58 billion additional dollars, is set to enable new advances in the next few years. Anthony and Andrew pointed out that only the IRS proved capable of catching infamous gangster Al Capone, while (tongue-in-cheek) acknowledging his negligence in limiting his financial operations to one jurisdiction. 

The next section of the conference invited TRACE partners to present key project results, with Dr Anastasia Kordoni from Trilateral Research opening with an overview of bias mitigation measures in an age of law enforcement AI. Technological and human safeguards are key to preventing bias, and the promotion of wider socio-technical understandings can reduce harms as new tools are developed and rolled out. Joachim Klerx from AIT then discussed the use of AI models to identify and structure networks of hidden relationships and knowledge that span the surface and dark webs. The TRACE knowledge graphs mark progress in this arena, with automatic reasoning the logical next step for those seeking to advance the field. The digital transformation of organised crime has seen crime-as-a-service reach new technical heights, with malware gangs, for example, woven into digital economies. Fresh resources and legislation are needed, and Professor Athina Sachoulidou offered legal evaluation; describing a transition from e-evidence into a world of AI-generated evidence, and the regulatory challenges that emerging technologies pose.

Thomas Havranek from CIN Consult took to the stage next, to demonstrate the TRACE toolkit itself and showcase its capabilities. With its user-friendly interface, AI analysis, and visualisation of mazes of transactions and entities, TRACE advances anti-money laundering technology according to in-depth consultation on practitioner needs.  Dr Dimitrios Kafteranis from Coventry University’s Centre for Financial and Corporate Integrity then spoke about the use of NFTs and crypto to shield assets, an issue that came to the fore during the lifetime of the project. This shift towards virtual assets was highlighted as a challenge by a range of speakers across the two days of the conference, with present regulation patchy and new European and US responses expected. The legal uncertainty is highlighted as counterproductive, given the need to protect the public from exploitation. 

Questions from the audience covered regulation, data acquisition, cryptocurrency, and the architecture of the TRACE tool. Partners spoke about the risks of AI for law enforcement applications, a recurrent theme of the project ‘s ethical, social and legal workstreams. Entity extraction and large language model functionality are among the technical achievements of the project, representing steps forward in the state of the art and attracting considerable audience interest. The TRACE consortium’s collaboration between technical, legal, and operational experts was cited as a strength, and a network that should be maintained formally or informally following the project’s conclusion to further bridge gaps in understanding. 

With the project headlines conveyed, the second conference keynote was delivered by Dr Niovi Vavoula, Associate Professor and Chair in Cyber Policy at the University of Luxembourg. Nina discussed the AI Act implications for law enforcement once again, with a focus on interoperability (i.e. how well systems work together). With the Act prohibiting certain AI systems, it is instructive to drill down into the detail. For example individual predictive policing is banned when based solely on profiling or the assessment of personality traits and characteristics (not where AI is supporting a human assessment), and the obligations of prohibited or high-risk applications shifted over the course of legislative deliberations. Interoperability is often regarded as a buzzword, but Niovi emphasised that the concept can be an enabler and driver of AI systems. 

A series of presentations explored the TRACE use cases and solutions in greater depth. This began with Lukáš Vilím and Martin Bartos from the Czech National Counterterrorism, Extremism and Cybercrime Agency. They offered the law enforcement perspective on participation with the TRACE project, with an overview of pressing issues in terrorist finance and other organised criminality from Russian gangs to the Satanist Order of Nine Angles. With cybercrime and hybrid threats a dominating public fear, the TRACE project provided a valuable opportunity for agencies such as this to lend their expertise and in turn benefit from research. Alison Schultz and Lucas Millan-Narotzky from the Tax Justice Network discussed regulatory arbitrage (i.e. organisations taking advantage of varying laws in different jurisdictions) and the geographic risk factors for money laundering. An entity moving its assets might consider the transparency obligations of real estate or high value asset registration, for example, linking back to the notion of tax haven vulnerabilities. The Tax Justice Network used statistical approaches to determine geographic risk, as summed up in an upcoming TRACE policy brief. They emphasise the need for legal and beneficial ownership registrations, without exceptions, limits on cash use and transportation, and coordinated international regulation of all implicated intermediaries. 

Thea Sogenbits from the Estonian Tax and Customs Board talked us through the processes and actors involved in contemporary credit card fraud, a crime that would be the third largest economy in the world if it were a country, before Professor Vladlena Bensonfrom Aston University provided a new angle on the value of AI for anti-money laundering. She reflected on technological advances over the duration of the TRACE project, and those of criminal operations during the same timespan. The costs of cybercrime exceeded USD seven trillion in 2022, and the issues of ransomware, malware, fraud and phishing have only grown more salient as AI has began to fuel them. There are big benefits in AI tools to counter these issues, but ominous pitfalls to avoid. Wrapping up day one of the conference, TRACE coordinator Professor Umut Turksen thanked our guests and speakers for their energy and insight, and guests filed off into the night for local delicacies at the Vienna State House.

Day two featured extended discussions on the future of money laundering, and AI in law enforcement. To open, Jan Ellermann from Europol delivered the third keynote. Jan offered the data protection practitioner’s view of these central issues, as well as the particular role of Europol in coordinating and directing information to local agencies. While Europol agents do not have direct enforcement powers, Jan noted that the agency had a dubious reputation amongst some of its counterparts for its emphasis on data protection safeguards. Reflecting on cases such as the dismantling of the EncroChat criminal communication system, Jan explained the value of data protection measures to train effective and ethical algorithms to refine otherwise unmanageable volumes of data (in the case of Encrochat, 120 million messages and 500,000 images). While privacy is vital, operational imperatives also require fast data protection processes to save lives in the most urgent situations. A positive data protection culture is thus good for everyone. 

In a brief intermission, Professor Turksen was proud to award the TRACE essay prize to Zhanin Fahd Al-Shargabi, a law student from Sofia University ‘St. Kliment Ohridski’, for her essay on the challenges of AI for European law and policing. Zhanin was awarded a €150 prize, a certificate, and the opportunity to present her findings to the conference audience. The essay is now available to read on our website here, and will be promoted as part of the final project newsletter. 

The fourth keynote saw Michael Nagl, an independent financial expert, provide insights on how private sector financial intelligence can be harnessed to reduce the risk of criminal transactions. Risk scoring involves assessing the frequency and size of payments, the products used, and the locations, histories and characteristics of the entities involved. These factors converge to indicate the possibility of money laundering or tax fraud, and organisations should be proactive in evaluating threats at the earliest juncture. 

The TRACE conference concluded with two panels. The first addressed the challenges of LEA uptake of new technology, with host Reinhard Kressl of VICESSE noting that too many European projects fail to deliver on their big ideas in a publicly impactful way.  The lineup included Tobias Mattes from the Bavarian Police, Peeter Paisuots from the Estonian Tax and Customs Board, Thomas Havranek from CIN, and Madison Leeson of the Istituto Italiano di Tecnologia and RITHMS project. Panellists debated how we can push adoption to frontline agencies, with resources such as the Europol Innovation Lab or Horizon Results Booster programme offering potential routes to the mainstreaming of new technologies.  They discussed challenges around legislation, political will, and intellectual property considerations and accreditations, complicating the smooth rollout of potentially revolutionary innovations into the marketplace. The exploitation of results is too often sidelined in research, with valuable work failing to reach its potential before a duplication of efforts as fresh consortia revisit familiar research subjects. 

The last panel took a different tack, focusing on the EU Anti-Money Laundering Authority’s new role in coordinating and empowering the activities of financial intelligence units, for example via shared templates for suspicious transaction reports. Discussion was led by Moran Harari from the Tax Justice Network, and featured Viktor Ivanov, Policy Officer at FISMA, Italo Borrelo, Staff Director at the Regulation and Institutional Cooperation Directorate of the Italian FIU, and Ololade Durodola from Coventry University’s Centre for Financial and Corporate Integrity. The panellists explored several areas of interest, from the remit of the AMLA to the functions of FIUs and banking countermeasures and the promises of advanced technology. Political agreement on the AMLA has been reached, with formal establishment set to conclude in the next few weeks. The potential of this new regime to reduce the harms of money laundering is considerable, and specialists and practitioners will watch its enactment with interest. 

And with that, the TRACE final conference was concluded. Reflections from Professor Turksen thanked participants and attendees for their contributions to an animated and enjoyable event, at the end of a project which assembled an energetic team of experts from varied disciplines and backgrounds to create a groundbreaking piece of technology. The remaining TRACE outputs, including its final policy briefs, closing video, newsletter, and reports will be released in the coming weeks and hosted on the website for future reference. Thank you, as always, for reading. 

Author: Dr Alex Murphy, Trilateral Research