British companies lost an estimated £600 million to AI-powered fraud last year, according to data published by the National Fraud Intelligence Bureau. The sharp rise has forced firms across London, Manchester, and Birmingham to accelerate spending on cybersecurity, reshaping how businesses protect their operations and customer data.
The Scale of the Problem
Authorities recorded more than 45,000 AI-assisted fraud cases in 2024, up from roughly 12,000 the previous year. Fraudsters now use generative AI to create convincing phishing emails, deepfake video calls, and synthetic identities at a scale that was impossible just three years ago. The speed and sophistication of these attacks have outpaced many companies' existing defence systems.
Small and medium-sized enterprises have been hit hardest. Trade body Make UK surveyed 500 firms in October and found that 68 percent had experienced at least one AI-related fraud attempt in the past six months. Of those, a third reported direct financial losses, with average damages around £45,000 per incident.
How Fraudsters Are Exploiting AI
The techniques used in current campaigns go beyond simple email phishing. Criminal networks are deploying AI chatbots to conduct extended conversations with victims, building trust before requesting money or sensitive information. Voice cloning technology now allows fraudsters to impersonate company executives in real time, convincing finance staff to authorise urgent transfers.
Deepfake video has emerged as a particularly costly tool. In one case reported by the City of London Police, a finance director in Leeds transferred £220,000 after a video call with someone he believed to be his chief executive. The caller used AI-generated visuals to mimic the executive's appearance and voice.
New Tactics Targeting Supply Chains
Beyond direct attacks on businesses, fraudsters are targeting supply chains with increasing frequency. By compromising smaller suppliers, criminals gain access to larger firms further up the chain. KPMG's cyber practice team documented 14 major supply chain attacks in the past year where AI tools played a central role.
The Economic Toll
Insurance broker Howden estimated that UK businesses will spend £8.7 billion on fraud prevention measures this year, a 31 percent increase from 2023. This includes new software licences, specialist staff hires, and third-party security audits. For many companies, the costs are squeezing budgets that were already under pressure from rising operating expenses.
The financial sector has been most aggressive in its response. Barclays, Lloyds, and HSBC have each announced multi-year programmes to upgrade fraud detection systems. The investments are partly defensive: insurers have raised premiums for cyber policies, making prevention cheaper than claims.
Regulatory Pressure Mounts
The Financial Conduct Authority warned in November that fraud now represents the single largest threat to consumer trust in digital banking services. The regulator expects lenders to implement stronger verification procedures by mid-2025 or face enforcement action.
Trade groups have pushed back, arguing that compliance timelines are unrealistic given the pace at which AI threats evolve. TechUK, representing more than 900 companies, told a parliamentary committee in December that firms need clearer guidance on what constitutes adequate protection under current rules.
Investment Flows Into Defence Startups
Venture capital has responded to the surge in demand. Cybersecurity firms specialising in AI threat detection raised £1.2 billion across 67 funding rounds in 2024, according to data from PitchBook. London-based startup Apiiro, which uses machine learning to identify fraudulent transactions, secured a £85 million Series C round in September led by Sequoia Capital.
Established players are also expanding through acquisition. BT Group announced plans to buy Bristol-based fraud detection firm FX3 for an undisclosed sum, aiming to integrate its tools across business customer accounts.
What Comes Next
Security researchers at the Alan Turing Institute expect AI-powered fraud attempts to double again in 2025 as tools become cheaper and more accessible. Law enforcement agencies face an uphill struggle: the National Crime Agency warned that fewer than one in ten cyber fraud cases results in prosecution, partly because criminal groups operate from jurisdictions with limited extradition agreements.
Companies that have not yet updated their fraud response plans should treat the coming months as a window to act. Several insurers now offer reduced premiums for businesses that complete certified training programmes for finance staff. The next major deadline comes in April, when new reporting requirements under the Online Fraud Charter require platforms to share more data with law enforcement.
