Limassol: For most of its existence the European Anti-Fraud Office has been associated with paper trails, forensic accountants and the slow reconstruction of where public money went missing. That image is now being deliberately overwritten. Between 22 and 24 June, OLAF and Cyprus’s Anti-Fraud Coordination Service gathered nearly a hundred specialists in this Mediterranean port city for the second Technical Workshop on Digital Anti-Fraud Tools, an event whose guest list said as much as its agenda. Investigators from the European Public Prosecutor’s Office, Europol, Eurojust and the European Court of Auditors sat alongside national fraud units to compare notes on a shared problem: the money is moving faster than the people chasing it.
The workshop’s substance was unapologetically technical. Sessions dwelt on open-source intelligence, the use of artificial intelligence to surface suspicious patterns in vast datasets, and satellite imagery to verify whether a subsidised road, farm or factory actually exists where a claimant says it does. Participants also presented the first results of the Digital Anti-Fraud Toolbox, an initiative launched in May last year to turn these techniques into standardised methods that any national authority can pick up rather than reinventing them case by case.
The urgency behind the gadgetry is structural. The Union now disburses money through instruments of unprecedented size and speed, from cohesion funds to the pandemic recovery facility, and every acceleration in spending widens the surface area for abuse. Traditional detection, which typically begins once the money is already gone, is poorly matched to programmes that pay out against milestones and self-declared results. The promise of the digital toolbox is to move detection earlier, flagging anomalies while a project is still live and recovery is still realistic.
Cyprus was a pointed choice of host. Earlier in June the island presented its first National Anti-Fraud Strategy at the fourth annual OLAF conference, part of a broader push by smaller member states to professionalise the machinery that guards EU money. For a bloc that has been criticised for uneven enforcement across its members, the sight of a national administration building its own capacity, rather than waiting for the centre to act, is precisely the direction OLAF wants to encourage.
Yet the enthusiasm for AI and satellites carries its own hazards. Automated pattern-matching produces false positives, and a system that flags honest recipients risks entangling legitimate projects in months of suspicion. Data protection lawyers note that trawling national and European databases to build a single investigative picture sits uneasily with the Union’s own privacy rules, and that evidence gathered by algorithm must still survive scrutiny in court. The tools, in other words, are only as good as the legal and human judgement wrapped around them.
There is also the awkward matter of jurisdiction. OLAF can investigate and recommend, but it cannot prosecute; that power now belongs to the European Public Prosecutor’s Office, which handles criminal cases in the participating member states. The two bodies have a working arrangement, but the division of labour remains a work in progress, and better detection only sharpens the question of who acts on what OLAF finds. A flood of well-evidenced referrals is of little use if the prosecutorial capacity to pursue them does not keep pace.
None of this was resolved in a three-day workshop, and it was not meant to be. The value of Limassol lay in signalling a direction: that the defence of the Union’s financial interests is becoming a data discipline, conducted in near real time, shared across borders and increasingly reliant on machines to see what auditors alone cannot. Whether that shift actually recovers more money, or merely generates more alerts, will be judged not in conference halls but in the recovery figures of the years ahead.
