Introduction
Along with many other parts of all our lives, artificial intelligence is transforming fraud. Technologies that improve efficiency for businesses and investigators are also enabling fraudsters to conduct more convincing scams at greater scale and lower cost. The age-old cat and mouse game between criminals and the authorities continues, super-charged by use of AI.
Generative AI allows criminals to produce realistic emails, messages, invoices and other documents within seconds. Deepfake technology enables the creation of synthetic audio and video capable of imitating executives, colleagues, professional advisers and public figures with increasing accuracy. These tools reduce the expertise traditionally required to commit fraud and increase the credibility of impersonation attacks.
AI-enabled fraud
Gone are the days when fraudsters relied on poorly drafted phishing emails, implausible lottery wins and the familiar promise of a fortune stranded in a foreign bank account. While those schemes were often undermined by obvious red flags, AI now enables criminals to produce communications that can be virtually indistinguishable from the real thing.
Government and industry evidence suggests that organised crime groups are already exploiting AI capabilities at scale. The Home Office’s National Assessment Centre Fraud Assessment 2025 found that fraud is increasingly online, technology-enabled and international, and that criminals are using generative AI and other fraud-enabling tools to scale attacks, lower barriers to entry and circumvent existing countermeasures.
The rise of deepfake fraud illustrates the scale of the challenge. Signicat reported a 2,137% increase in detected deepfake fraud attempts over a three-year period and found that AI-related attacks now account for more than 40% of detected identity fraud in the financial sector. Regula’s Deepfake Trends 2024 survey similarly found that approximately 49% of organisations had experienced audio or video deepfake fraud.
AI increases fraud volumes and undermines traditional methods of verification in parallel. Poor grammar, unusual wording and obvious inconsistencies were once common indicators of fraud. AI-generated communications are often indistinguishable from legitimate correspondence, making it harder for businesses and individuals to identify fraudulent activity before losses occur.
Legal response
From a legal perspective, English law has generally adapted well to technological change, though practical and resourcing challenges remain.
The Fraud Act 2006 remains capable of addressing many forms of AI-enabled deception. Whether a fraudulent misrepresentation is made directly by an individual or generated through AI, criminal liability will generally arise where the statutory elements of the offence are satisfied. From a civil recovery perspective, AI changes little in principle. Established civil causes of action, including deceit, conspiracy, dishonest assistance and knowing receipt, also remain available to victims seeking recovery.
Recent legislative developments have strengthened that framework. The Economic Crime and Corporate Transparency Act 2023 introduced the new corporate offence of failure to prevent fraud, which came into force in September 2025. The offence expands potential corporate liability where fraud is committed by associated persons for the benefit of the organisation.
The law has also evolved in response to the increasing use of digital assets in fraud. A poignant example of the English common law adapting to new technologies can be found in the much discussed case of D’Aloia v Persons Unknown, where the High Court confirmed that cryptoassets are capable of being traced and subjected to proprietary remedies, reinforcing the tools available to victims seeking asset recovery.
The Property (Digital Assets etc.) Act 2025 subsequently placed that position on a statutory footing by confirming that digital assets can attract property rights under English law, which further strengthened the basis for proprietary claims and tracing remedies involving cryptoassets.
Challenges in enforcement and recovery
However, despite a stronger legislative arsenal, enforcement presents challenges.
To return to the Home Office’s Fraud Assessment 2025, fraud affecting the UK is increasingly international, technology-enabled and difficult to investigate. Fraudsters can operate more easily from overseas jurisdictions, use AI-generated identities and forged documentation to conceal their involvement, and move stolen funds through opaque corporate structures and cryptocurrency wallets within hours. As noted in Fraud, Asset Tracing & Recovery 2026: England & Wales, assets in high-value fraud cases “rarely sit transparently in the defendant’s personal name” and may instead be held through “nominee shareholders, discretionary trusts [and] layered holding companies”. Assets may also be owned through entities incorporated in offshore financial centres, including the BVI, Cayman Islands and UAE free zones, creating additional jurisdictional and enforcement challenges. Recovery may require parallel proceedings in more than one jurisdiction, disclosure applications against trustees, registered agents or exchanges, and coordination across multiple legal systems.
Identifying perpetrators is also becoming more difficult as deepfakes and AI-generated documents obscure who is behind a fraudulent communication or transaction. Tracing assets presents similar difficulties, particularly where cryptoassets are moved through multiple blockchain wallets and custodial accounts. Even where the legal status of digital assets is clear, claimants can still face evidential challenges in proving that specific assets passed through a particular wallet or exchange. As the CDR chapter observes, successful recovery ultimately depends on legal remedies and the ability to secure timely disclosure, trace beneficial ownership, freeze assets and enforce orders across jurisdictions.
Fighting fire with fire
The response to AI-enabled fraud is technological as well as legal.
Financial institutions are deploying AI-driven monitoring systems capable of analysing vast volumes of transaction data and identifying anomalous activity in real time. HSBC, for example, has reported using AI systems to analyse approximately 900 million transactions each month as part of its fraud detection processes.
In many respects, the fight against modern fraud has become an AI arms race, with AI being deployed against itself. As fraudsters increasingly rely on AI to enhance the scale, speed and sophistication of their schemes, organisations are turning to AI-driven detection and investigative tools to counter those threats.
For example, there is growing use of AI-assisted investigations, technology-assisted review and blockchain analytics in civil fraud and asset recovery proceedings. These technologies are often being used to combat frauds that have themselves been enhanced by AI, from deepfake impersonation scams and AI-generated phishing campaigns to the rapid movement of stolen cryptoassets through complex transaction networks. AI-powered tools can help practitioners analyse large volumes of financial data, identify suspicious patterns, trace assets across jurisdictions and support disclosure exercises more efficiently than would be possible through manual review alone.
In practice, businesses are increasingly required to match the technological sophistication of fraudsters. As AI becomes embedded in both the commission and detection of fraud, utilising comparable technologies is no longer a matter of choice but a practical necessity.
Conclusion
AI is reshaping the fraud landscape, but the greatest challenge lies in enforcement. AI has increased the volume of fraud and narrowed the window in which victims can respond. Fraudsters can exploit automated systems, synthetic identities and digital assets to move value at extraordinary speed, often leaving investigators to reconstruct transactions that span multiple jurisdictions and intermediaries.
Although English law provides powerful remedies, they are only as effective as the technology available to support them. Success therefore depends increasingly on rapid investigation, effective asset tracing and the ability to obtain actionable intelligence before assets are dissipated. As AI-enabled fraud continues to evolve, those responsible for preventing, investigating and enforcing legal rights must become increasingly fluent in the technologies and behaviours they seek to address. In an environment shaped by AI, effective fraud prevention and asset recovery will depend on effective anticipation rather than reaction.
