Ted Datta, Senior Director and Head of Industry Practice for Europe and Africa at Moody's, spoke at Counter Fraud 2026 about the new legal playing field created by the Economic Crime and Corporate Transparency Act (ECCTA), the continued abuse of UK company formation, AI adoption among compliance professionals, and how early warning signals can be used to detect fraud before it takes hold. Here is a summary of what he said:
Ted opened with a case study - Sharon, a 68-year-old retiree who was drawn into a task-based commission scam and then a romance fraud, losing money she could not afford to lose. Her case was one of 283 reports, totalling £7.4 million of fraud in a single year in one regional policing area. Ted asked the room to hold onto three statistics: 300 victims of self-harm and suicide linked to fraud, £17.5 million lost to pension fraud, and an average loss of £8,000 per victim of romance scams in the 75-84 age group. We are, he said, living through a fraud epidemic, which targets increasingly vulnerable people.
Ted gave a detailed update on Companies House reform. The changes since 2024 have been meaningful: a new statement of lawful business purpose, a new appropriate address regime, the introduction of Authorised Company Service Providers (ACSPs) - of which 1,361 are now certified - and identity verification for new Directors and Persons of Significant Control (PSCs). The next step, under way this year, is rolling that verification requirement out to all existing companies as they submit confirmation statements.
Moody's tracked the impact of these changes across seven shell company indicators in their Orbis database. The trend is encouraging. Companies formed with high-risk jurisdictional ownership have dropped from around 6,000 a month in 2024 to 1,500 in the last six months. Outline directorships - where a Director holds an unusually high number of roles - are down approximately 45%. Mass registrations, where companies are formed simultaneously at the same location with similar names, are also falling.
But they are not gone. Around 4,000 mass-registration companies are still being formed every month. Moody's built a high-density address indicator to identify specific postcodes where large numbers of companies are registered without any legitimate service provider at that address. One example: 268 companies registered to a fast food restaurant in Weston-super-Mare. One of those companies was a money service business linked to pig-butchering cryptocurrency scams. The bottom line, Ted said, is that Companies House is on a journey, but fraudsters are still taking advantage.
On the Failure to Prevent Fraud (FtP) offence, Ted highlighted two aspects that he sees generating the most questions: First, the breadth of who counts as an associated person - not just employees, but agents, service providers and suppliers. If they commit a fraud offence in the course of providing services to your organisation, you may be liable. Second, the extraterritorial reach: if the victim of a fraud is in the UK, an overseas business could find itself in scope, even if it has no UK presence. The Serious Fraud Office guidance on adequate procedures is, he said, a meaningful step forward - with specific reference to the use of appropriate technology for due diligence, including third-party screening tools and checks on trading history and professional status. Inaction, he was clear, is no longer a defensible position.
Moody's surveyed 650 risk and compliance professionals on AI adoption. In 2023, 9% were actively using AI in their risk and compliance function. By 2025, that had risen to 24%, with a further 29% trialling or piloting it - meaning 53% are now actively engaged with the technology. A show of hands in the room at the Counter Fraud 2026 conference produced, Ted noted, exactly the same figure.
96% of Moody's survey respondents said AI would impact their role as it becomes embedded in day-to-day operations. The top perceived benefit - selected by 70% - was reducing repetitive manual tasks. Other frequently cited advantages included handling complex scenarios, processing large volumes of data, and accelerating decision-making. On the question of how roles will change, 61% expect to take on more advisory and strategic work, 54% expect to spend more time developing and fine-tuning AI tools, and 47% see themselves acting as supervisors of AI systems.
The most important finding, he said, was the correlation between data quality and AI effectiveness. Organisations with superior data quality were two and a half times more likely to be actively using AI. Large Language Models (LLMs) can structure information in ways that were not previously possible - but they cannot fix poor underlying data. That is a problem organisations need to solve themselves, before they can expect AI to deliver.
Ted walked through a real case - Stenn, a fintech invoice financing company with a £900 million valuation that collapsed into administration after HSBC called in its revolving credit facility. Investigators found that payments had not been going to the named blue-chip companies in the invoices, but to shell companies in Serbia with no connection to those businesses. All four of what Ted called the four early warning signals were present:
He demonstrated how an agentic AI tool - grounded in Moody's proprietary data covering 608 million companies, 427 million individuals, and 20 million adverse media profiles - could surface all four signals in minutes, producing a structured due diligence report that would previously have taken days of Analyst time.
Ted closed with four direct conclusions: