AI is changing embedded finance in two distinct directions at once.
Through embedded finance – which is the integration of payments systems into non-financial websites, platforms or apps – payment making is faster and smarter, and exists within digital journeys.
However, embedded finance also gives fraudsters cheaper tools, better disguises and more scalable ways to attack. AI is core to embedded finance, providing greater efficiency, UX, and personalisation, but it’s also an aspect where the risks are still emerging.
The real question is no longer whether AI should be adopted, but how it can be deployed without automating risk alongside growth.
Fraudsters aren’t slowing down. The research tells us fraudulent activity is only increasing. UK Finance says criminals stole £1.17 billion in 2024, while Cifas recorded more than 444,000 fraud cases in 2025, the highest annual total on record. Fraud is not a side issue for payments providers; it is a direct challenge to trust and customer retention. The opportunity, however, is just as significant.
Payments, technology, and customer-centric service
The latest Bank of England and FCA survey shows that 75 per cent of financial services firms are already using AI, with another 10 per cent planning to do so in the next three years. It also illustrated that the highest current benefits are in data and analytical insight, anti-money laundering and combatting fraud, and cyber security.
While AI has certainly captured the imagination of businesses looking to transform, I’d argue the risks and opportunities are not isolated to AI innovation, but data-centric technological progress as a whole. The payments sector needs to think of innovation as not only a way to expand services across onboarding, transaction monitoring, reconciliation, servicing, compliance and disputes, but to harness the true power of data.
For providers, a combination of technologies, such as AI and ML, can improve far more than fraud controls. It can streamline onboarding, strengthen transaction monitoring, personalise customer support, reduce false positives and help firms design new products around real patterns of customer behaviour.
Yes, correct implementation of technology means businesses can launch more sophisticated products and handle the workloads associated with them more efficiently. But it can also help ochestrate vast amounts of data.
The AI risk – payments fraud is on the rise
While payments innovation brings opportunities for business, the same technologies are expanding the ventures of fraudsters and criminal organisations.
UK Finance’s report found that 70 per cent of authorised push payment (APP) fraud cases began on online platforms. In practice, AI is already making impersonation, phishing, synthetic identity creation and account takeover more convincing and more efficient. In a world of instant onboarding and embedded checkout, a weak control can be exploited at machine speed.
What exists now is a landscape where fraudsters are constantly evolving their tactics and targets. Victims are unprepared and vulnerable to emerging threats. It is not the fault of the consumer. They should rightly trust that easier, more efficient payment making is secure. However, they remain unaware of how common APP fraud has become, as AI helps fraudsters scale the volume and sophistication of attacks.
BoE research shows only 34 per cent of firms say they have a complete understanding of the AI they use, while 46 per cent report only a partial understanding. It’s this gap in knowledge that can be exposed by AI-wielding fraudsters. Consumer protections, particularly in the UK with its strong regulatory approach to fintech services, means successful reimbursement in the event of fraud is likely; the Payments Systems Regulator reports 84% of consumers are reimbursed within five days.
It’s not the case for businesses, with fraud losses an accepted part of operations – according to the National Crime Agency, 86% of business fraud goes unreported.
So, in order to harness the power of AI technology, the payments industry must start creating a framework that protects businesses from fraud while also safeguarding the customer to ever great extent.
This reality would not be a momentous feat.
Innovation with collaboration
Businesses already harness their data powerfully, albeit we must recognise that AI could aid in connecting fragmented signals to spot anomalies earlier. Yet it is not the connection of separate systems within a business that is the main hurdle.
In my view, the requirement is collaboration.
UK Finance has argued that tackling fraud requires the public and private sectors to use data and intelligence more effectively, while the Bank of England is building its AI Consortium to support dialogue on the safe use of AI in financial services. Payments providers, banks, platforms, telecoms firms, merchants and regulators all see different parts of the same threat. If those insights stay disconnected, criminals keep the advantage.
AI will absolutely expand what embedded finance can do. It can improve customer care, sharpen data management, widen product innovation and strengthen fraud defences.
However, in order to build trust and security across embedded finance, financial services need to use the collective might of our data, fraud professionals, and AI advantages, to unite against fraudsters, regardless of whether tactics are legacy or AI-enhanced.
It is my view that, only then, can we be sure our innovations are made on sure footing.
The time to implement these systems and structures is now.
