Digital innovation and the challenges of enforcement

The authority of law enforcement agencies is tied to the results they achieve. With increasing digitization and globalization, these agencies face the challenge of keeping pace with such developments. To that extent, authorities and their authority are also vulnerable to these factors.

Crime has partly shifted to the digital realm, and internationalization means that criminals can more easily carry out activities and move assets abroad. However, the sword cuts both ways. At the same time, digitization and international cooperation provide authorities with the tools and opportunities to investigate and enforce the law more efficiently and effectively. The question is how authorities can adopt a more future-oriented approach and become less vulnerable to new technology.

Some examples

Three recent examples illustrate the pressure that enforcement authorities are currently under. The first is cybercrime. On 16 March 2026, an Interpol report warned of an increasingly sophisticated global financial fraud threat [1]. The threat assessment points to hybrid tactics, global growth in scam centres and a rise in sextortion. The second is confiscation. In recent years, the proceeds of confiscating criminal assets have been disappointing, due to the absence of large settlements. The confiscation estimate is therefore revised down structurally by 284 million euros [2]. The third is healthcare fraud. Tax money spent on care rises from €107.4 billion in 2019 to an expected €202 billion in 2050.Money that ends up in criminal pockets: €12 billion [3]. The chance of being caught is very small.

Some factors that make enforcement more difficult

These examples are not isolated incidents. They result from three underlying factors that make enforcement structurally more difficult. First, digital identities and digital tools, which bring anonymity and execution power. Offenders can mask their identity online while leveraging powerful digital tools — encryption, automation, and dark-webinfrastructure — giving them capabilities that outpace traditional investigative methods. Second, government systems are still partly based on a high-trust society, which creates structural vulnerability and less incentive. Many governmental systems were built on assumptions of identifiable actors, mutual trust and physical presence. These foundations erode when actors exploit the trust built into the system. Third, complex legal structures, internationalisation and fast movement of assets, combining corporate structures with speed. Layered corporate structures, offshore vehicles, and instant cross-border transfers allow assets to be shielded off and move faster across borders. The implication for enforcement: when digital complexity, structural vulnerability and internationalisation are combined, this can undermine institutional strength — the legitimacy and effectiveness of enforcement authorities are tied to the robustness of the very systems they protect.

Three possible responses

How should authorities respond to these developments? Broadly speaking, there are three options. The first is to restore societal trust, so that fewer checks are needed. The question is whether this is realistic, or whether digital complexity, combined with the other factors, makes systems based on trust inherently fragile. The second is to continue with selective enforcement: intervening in a limited number of cases in order to deter others, and accepting the limitations of this approach. The third is to invest in people and technology. This means scaling up investigative capacity through specialist units, AI-assisted analysis and cross-agency collaboration, in order to match the volume and speed of digital crime.

Easier said than done

Enforcement of parking and speeding offences is already largely automated. This raises the question of whether AI will, in time, also take over the enforcement of simple to intermediate violations and offences. Consider a selection of AI branches. LLMs, for knowledge and content production: models that generate, summarise and reason over text, enabling report drafting, legal research and pattern recognition. Data-driven predictive analytics: models that identify risk patterns, flagging suspicious transactions and prioritising enforcement targets. And automation through agents: AI agents that execute multi-step enforcement workflows, from data collection to case preparation.

How might AI help overcome digital complexity?

AI can support law enforcement in three ways:through automation, detection and analysis. First, automation. Routine tasks such as data collection, report drafting and case preparation can increasingly be handled byAI systems and agents. This frees up capacity for the cases that require human judgement, and allows enforcement to keep pace with the volume and speed of digital crime. Second, detection. Predictive models and continuous monitoring can identify patterns of abuse, suspicious transactions and violations as they emerge, rather than years later. This helps authorities prioritise their efforts and strengthens the resilience of government systems. Third, analysis. By combining public and non-public data, AI can help investigators pierce the anonymity of online offenders and the layers of complex corporate structures, two of the main obstacles enforcement faces today.

Challenges

The use of AI in enforcement is not without its problems. The most obvious one is privacy. AI systems that process personal data must respect fundamental rights, which requires a lawful basis, proportionality and purpose limitation. These principles sit uneasily with broad surveillance. Then there is the risk of error. Automated systems will flag innocent people and miss guilty ones. How much of each is acceptable depends on what is at stake.

More effective detection may also change how cases end. If AI substantially increases detection rates, more cases may be resolved through settlements rather than prosecution. What that means for deterrence and justice is an open question. AI also blurs the line between supervision and criminal investigation. Continuous monitoring makes it harder to determine when supervision turns into an investigation, and therefore when legal safeguards should apply. Finally, combining data across agencies and borders remains legally and technically complex. Incompatible systems, data silos and conflicting legal bases all slow down AI-assisted enforcement.

Conclusion

Digitisation and internationalisation have made enforcement authorities more vulnerable. Crime moves faster, hides better and crosses borders more easily than the systems designed to address it, as the examples of cybercrime, confiscation and healthcare fraud show.

Yet the same technology can also strengthen enforcement. AI can automate routine work, detect abuse as it emerges and help investigators see through digital and legal identities. At the same time, it raises fundamental questions about privacy, errors, settlements and the boundary between supervision and criminal investigation. The question, therefore, is not whether to use AI, but how. Authorities that invest in technology and expertise, while safeguarding the legal principles on which their work rests, can become less vulnerable without losing the legitimacy that gives enforcement its authority.

[1] Interpol,“Interpol Report Warns of Increasingly Sophisticated Global Financial FraudThreat,” 2026, https://www.interpol.int/News-and-Events/News/2026/INTERPOL-report-warns-of-increasingly-sophisticated-global-financial-fraud-threat

[2] Ministry of Finance, Spring Memorandum 2026 (Voorjaarsnota 2026), 2026, https://www.rijksfinancien.nl/voorjaarsnota/2026

[3] RegionalInformation and Expertise Centre East Netherlands (RIEC Oost-Nederland),Multi-Year Policy Plan on Healthcare-Related Crime 2025–2026. https://www.vrtwente.nl/sites/vrt/files/vergaderstukken/E2a%20meerjarenplan%20RIEC%20ON%20zorgcriminaliteit.pdf

‍