RegTech and fraud prevention.

The panel

SGR Compliance took part as a partner in the 2025 edition of Forum Banca, the event dedicated to innovation in the banking and financial sector.
Throughout the day, banks, institutions, fintech companies, and RegTech providers shared insights on the trends reshaping the industry. The discussion spanned from the evolution of business models to the digitalisation of processes, and the new frontiers of compliance and financial crime prevention.

The 2025 edition placed the AML debate at the centre of discussion, focusing on how artificial intelligence and RegTech technologies can make anti-money laundering controls and fraud prevention more effective, enabling the shift from reactive to predictive models.

SGR took part in the panel “How AI and RegTech are transforming AML models towards a proactive approach to fraud prevention,” moderated by Simone Mazzonetto, Adjunct Professor of Anti-Money Laundering at Ca’ Foscari University of Venice and President of AML LAB.

The discussion featured leading professionals from the financial and compliance sectors:

• Andrea Danielli, CEO, Mopso;
• Valentina Gilberti, Head of Customer Success & Industry Insights, SGR Compliance;
• Marco Valcavi, AML Officer and MLRO, Banca Mediolanum;
• Salvatore Marrone, Group Head of Compliance & AML, Banca Sistema;
• Michele Pisani, Chief AML Officer, BPER Banca;
• and Marco Paudice, Head of Delivery, Joint Services.

1. Connecting data to detect anomalies

Addressing the first question — “How can AI and blockchain strengthen AML controls by anticipating fraud and anomalies in real time?” — Andrea Danielli illustrated the various ways AI can be applied to anti-fraud efforts, stressing that its effectiveness depends on the ability to read and interpret large volumes of heterogeneous data.

Building on this, Valentina Gilberti from SGR Compliance highlighted the science of data connection: the real innovation, she explained, lies not only in the power of AI but in the structure and quality of the datapoints that feed it.

Hence the importance of structured, historicised datasets complete with unique identifiers, contextual metadata, and explicit relationships between individuals, legal entities, and transactions.
On this basis, graph analytics and network detection can uncover risk patterns and suspicious behaviours, reducing false positives and improving the timeliness of alerts.

“RegTech and AI solutions deliver real value when data is organised, enriched, and designed to create meaningful connections. A single news item or isolated event isn’t enough — it’s about reconstructing context and identifying risk patterns that would otherwise remain hidden.”

2. The evolving role of the AML Officer

The second question — “How is the role of AML Officers changing with the adoption of RegTech and automation tools?” — featured insights from Marco Valcavi (Banca Mediolanum) and Salvatore Marrone (Banca Sistema).

Both emphasised how the AML profession is undergoing a major evolution: alongside legal expertise, technical and analytical skills are becoming increasingly essential. Automation platforms and machine learning systems do not replace human judgment but demand greater capacity to interpret, contextualise, and validate algorithmic outputs.

Collaboration between human intelligence and technology is therefore key to ensuring effective controls and well-founded decisions.

3. Digital models and intelligent automation

Responding to the third question — “Which digital models can overcome the limits of traditional compliance, aiming for proactivity and operational efficiency?” — Michele Pisani (BPER Banca) and Marco Paudice (Joint Services) explained how advanced digital frameworks integrate automation within compliance processes without reducing the importance of human oversight.

The goal is not to replace the AML expert but to free them from repetitive activities, enabling focus on complex case analysis and preventive strategies. Technology thus acts as a multiplier of effectiveness, combining accuracy, speed, and cost efficiency.

Deep Dive

Datapoints as the engine of proactivity

RegTech, machine learning, and similar technologies go beyond automating controls: they enable institutions to anticipate risks by identifying connections, patterns, and behaviours before they materialise as fraud or AML violations.

Their effectiveness depends on the quality, timeliness, and structure of the data powering the entire compliance ecosystem. In her intervention, Valentina Gilberti explored the role of data as the foundation for a predictive and timely compliance model, built on two key dimensions:

1. Connections
2. Timeliness

1. Connection — Linking data to identify risk patterns

Technology can decisively enhance data management and analysis — but only when data are qualitative from the start, beginning with the onboarding phase.

Gilberti explained that RegTech solutions reach their full potential only when fuelled by organised, verified, and contextually enriched data.

She pointed out the limits of pure “web crawling”, used by some providers to scrape large amounts of information from the internet: such data are often fragmented and lack context, leaving users to interpret, correlate, and verify them manually. This approach provides only a static, partial snapshot of risk and captures data only from indexed sources, overlooking potentially relevant information.

> Over 20 years of data

By contrast, an effective AML methodology transforms informational events into structured datapoints with unique identifiers, historical context, and explicit relationships. In this context, SGR Compliance stands out for its proprietary database developed over more than twenty years, built on multiple informational layers that ensure depth, consistency, and precision in data attribution.
This architecture makes it possible to connect information from heterogeneous sources, reconstruct the evolution of entities over time, and provide algorithms with a solid, contextualised, and verifiable data foundation — essential for making AML controls truly predictive and effective.

> Human expertise

In the data selection and validation process, the integration of human intelligence ensures a qualitative layer of control that automated technologies alone cannot guarantee. Analysts assess relevance, reliability, and consistency of sources, identify inconsistencies, and remove irrelevant data.

This human intervention is critical to validate the links generated by entity resolution systems, ensuring that the final dataset is verified, contextualised, and operationally useful for AML and anti-fraud processes.
“Entity resolution” makes it possible to match and reconcile information referring to the same subject, even when it originates from different sources or time periods. This prevents duplication and ambiguity caused by name similarities or incomplete data, reduces false positives, and reconstructs a coherent, continuous, and verifiable informational biography of the subject.

> Applications

Thanks to verified, structured, and interconnected data, technological solutions can be applied across the entire AML control lifecycle — from onboarding to continuous monitoring and case management:

• During onboarding (KYC/KYB): automated data reconciliation, cross-validation across AML, commercial, and public sources, and biometric analysis to detect synthetic identities or deepfakes.
• In real-time screening and ongoing monitoring: automated checks on sanctions lists, PEPs, and adverse media, with live updates (e.g., sanctions within 60 minutes).
• Through graph analytics: mapping and analysing relationships among individuals, legal entities, assets, and transactions to expose opaque structures such as shell companies, nominees, or trusts.
• In complex corporate networks, this approach highlights critical nodes and indirect relationships, revealing participation or beneficial ownership schemes that may appear harmless in isolation but form anomalous or illicit configurations when viewed as a whole.

2. Timeliness — Detecting hidden risks and enabling rapid decisions

The second key dimension is speed.

A dataset may be technically accurate but loses value if updated too late.

Modern compliance increasingly relies on near real-time data feeds, automated workflows, and predictive tools that can anticipate emerging risks.

Gilberti emphasised that data timeliness is now a fundamental condition for dynamic, preventive compliance.
For instance, SGR’s Sanctions Live Screening module updates sanctions data within 60 minutes of official publication, dramatically reducing exposure and ensuring compliance with the EU Regulation 886/2024 on instant payments, which mandates continuous screening of payment service users.

But timeliness extends well beyond sanctions.

Technology now makes it possible to:

• Apply predictive models that adapt automatically to new risk patterns (adaptive learning).
• Automate escalation and alert management, improving efficiency and reducing false positives.
• Support rapid and transparent decision-making, with predictive dashboards that help compliance teams understand alert logic and prioritise cases.
• Integrate predictive intelligence models, such as early warning systems and risk clustering by geography or sector, to anticipate emerging trends.

Timeliness thus becomes not just a matter of fast updates but a decision-making enabler. Ultimately, timeliness translates into an ability to respond swiftly and with full awareness, supported by predictive analysis, intelligent automation, and data-driven decision-making.

Key Takeaways

• Data are the true engine of predictive compliance — from onboarding to the continuous enrichment of datapoints over time.
• Data quality — unique identifiers, historical depth, and explicit relationships — defines the accuracy of AI models in distinguishing real risk from noise.
• Timely updates (e.g., sanctions within 60 minutes) reduce exposure and enable immediate responses.
• The combination of advanced technologies and human intelligence transforms AML controls from reactive to proactive, enhancing precision, efficiency, and explainability.
• True innovation in compliance is not merely technological but informational: building structured, timely, and verifiable knowledge is the key to anticipating risk.