UK Trends and Developments Contributed by: Phillip D’Costa, Richard Marshall, Sophie Newman and Harriet Campbell, Penningtons Manches Cooper LLP
been direct physical interference with the cold wallet in which Bitcoin was held. It remains to be seen whether a tort of wrongful inter - ference with digital property will develop, or whether a cause of action in conversion will apply to intangible assets. This case, coming just months after the Digital Assets Act came into force, has already stretched the boundaries of the law as it applies to digital assets. It seems likely that further pressure will be placed on traditional property torts as the fight against fraud evolves. Artificial Intelligence For banks, PSPs and fintechs, AI is reshaping the fraud landscape in causing significant financial, operational and reputational harm. From hyper-realistic deepfakes to phishing emails generated instantaneously and tai - lored precisely to a company’s structure and tone, AI is industrialising deception on a grand scale, enabling fraudsters to sharpen their targeting in real time and scale attacks on the financial industry. In August 2025, Anthropic reported in its “Detecting and countering misuse of AI” announcement several examples of Claude being misused by fraudsters, including a large-scale extortion operation using Claude Code. Key findings included the following: • Agentic AI has been weaponised. AI models are now being used to perform sophisticated cyber - attacks rather than just advising on how to carry them out. • AI has lowered the barriers to sophisticated cyber - crime. Fraudsters with few technical skills are using AI to conduct complex operations, such as devel - oping ransomware, that previously would have required significant training. • Cybercriminals and fraudsters have embedded AI throughout all stages of their operations – includ - ing profiling victims, analysing stolen data, stealing credit card information and creating false identities, allowing the expansion of the reach of their fraud operations to more potential victims. Following its initial announcement, in November 2025, Anthropic shared in its report titled “Disrupting the first reported AI-orchestrated cyber espionage campaign”
the steps it had taken to build stronger safeguards, offering suggestions for detecting, disrupting and pre - paring for future versions of this type of attack. Similar reports were produced by Google (“Adversarial mis - use of generative AI”) and Open AI (“Disrupting mali - cious uses of AI”). From a legal perspective, financial institutions need to be prepared not only to combat AI cybercrime but also to meet the legal and regulatory requirements in all jurisdictions in which they operate. Companies need to understand and monitor their obligations (and liabilities) in a range of regulatory regimes. This includes cyber insurance, internal AI governance and misuse detection, as well as dealing with data and privacy obligations at scale and speed. Key issues include the following: • In England and Wales, failure to take appropriate measures to guard against AI-enabled fraud may lead to large organisations facing liability under the Failure to Prevent Fraud offence, introduced by the Economic Crime and Corporate Transparency Act 2023. • In the EU, under the EU AI Act (Regulation (EU) 2024/1689), organisations are obliged to ensure that AI systems are designed and monitored to prevent unlawful outcomes, including, for example, risk assessments (also a requirement for compli - ance under the UK Failure to Prevent Fraud obliga - tion). • Within the EU again, under new rules building on the EU Digital Services Act, social media compa - nies operating in the EU will be required to com - pensate banks by covering their customers’ losses if it is clear that a user has been defrauded as a result of the platform’s failure to remove a reported scam. While banks must also refund customers for certain unauthorised transactions, the two-pronged approach to liability is intended to avoid consumer claims falling between the cracks. In terms of future obligation, there are calls from some financial institutions for the net to be spread more widely for consumer compensation where, for instance, the fraud itself has originated from – and was arguably enabled by – a social media platform. Data produced by UK Finance suggests that 72% of APP
373 CHAMBERS.COM
Powered by FlippingBook