- More than 75 per cent of City firms already use AI, according to a UK Treasury select committee report.
- Moody's warns that an outage at one major AI provider could potentially affect banks and other sectors at the same time.
- Banks could also face higher AI costs, data risks and faster movement of customer deposits between accounts.
Banks are rushing to bring artificial intelligence into everything from customer service to credit decisions, but Moody's is warning that the technology could create a new weakness for the financial system: too much dependence on too few technology companies.
The rating agency says the growing use of AI by banks and insurers could eventually reduce costs and increase revenues. But as financial firms adopt similar systems, they are also becoming increasingly reliant on a relatively small group of companies providing AI models and cloud computing infrastructure.
That creates what Moody's calls "systemic dependency". In simple terms, if a major AI or cloud provider suffers a serious outage, the problem could potentially spread across several banks and financial institutions at once.
This is different from a bank's traditional technology risk. A bank might normally have several suppliers supporting different systems. With generative AI, however, many firms could end up relying on the same foundation models, cloud infrastructure and technology providers.
Moody's said, as quoted in a news report, that an outage at one major provider could spread quickly across customers and sectors. As AI adoption deepens, regulators could therefore pay more attention to operational resilience and the concentration of third-party AI providers.
The concern is not hypothetical in the sense that banks are only experimenting with AI. More than 75 per cent of City companies were already using AI, according to a UK Treasury select committee report published in January. Insurers and international banks were among the biggest users.
Much of that technology is being used for relatively routine work, such as administrative tasks. But AI is also moving into more important parts of financial operations, including processing insurance claims and assessing customers' creditworthiness.
That means a technology failure could potentially affect much more than a chatbot.
There is also a question of price.
Could AI suppliers gain the upper hand?
Moody's has identified another risk it calls "vendor dependence". If a small number of AI model and infrastructure companies become dominant, they could potentially gain greater influence over the prices banks pay for their technology.
That matters because banks are spending heavily to keep pace with competitors. The rating agency expects AI adoption to require substantial investment, while warning that competition between financial firms could eventually eat into some of the financial benefits.
In other words, if every major bank spends heavily on the same technology to achieve similar efficiencies, those gains may become less of a competitive advantage and more of a basic cost of doing business.
The pressure could become greater if AI companies themselves come under pressure to generate profits from the huge sums being invested in the sector. Moody's has pointed to companies such as OpenAI and Anthropic as examples of major AI developers facing investor expectations around profitability.
For banks, that could eventually mean higher prices for access to AI models or computing capacity if a handful of providers gain enough market power.
There are ways for financial firms to reduce that exposure. Moody's noted that banks and insurers have experience negotiating technology contracts and could use open-source AI models or build partnerships with multiple providers.
Banks also retain control over important assets such as their proprietary data, according to the rating agency. That could give them some protection if they avoid becoming locked into a single technology supplier.
But AI creates other risks at the same time.
Greater use of the technology could increase concerns around data privacy, cyber security and fraud. Banks also need to consider what happens if AI systems make decisions that are difficult to explain or if a model behaves differently after an update.
There is another issue that is particularly relevant to banks: deposit flight.
AI could make it much easier for customers to compare accounts and move their money towards banks offering higher interest rates. If switching becomes increasingly automated, large amounts of deposits could potentially move between institutions much faster.
For banks, deposits are a key source of funding. A sudden loss of deposits can therefore create pressure on liquidity and confidence.
Moody's said depositors' trust in their bank, together with the resilience and stability of deposit funding, would remain critical as AI makes financial services more responsive.
The technology is also likely to change the workforce. Moody's estimates there is a 20 per cent chance that AI could perform the work of a solid mid-level employee by 2030. That does not mean one in five banking jobs will disappear, but it underlines the uncertainty around how much human work advanced AI could eventually handle.
Lloyds Banking Group offers a current example of how large that transformation could become. Chief executive Charlie Nunn has backed a £13 billion strategy built partly around AI, with the bank targeting £2 billion in cost cuts. Nunn has acknowledged that the programme will affect work and require staff to be reskilled while the bank also recruits new technology specialists.
So the warning from Moody's is not that banks should stop using AI. The financial sector appears increasingly committed to the technology because of its potential to reduce costs, improve processes and generate new revenue.
The bigger question is who controls the technology underneath it.
If dozens of banks eventually rely on the same handful of AI models and cloud systems, an outage, cyberattack, pricing change or major technical failure at one provider could have consequences far beyond that individual technology company.
For regulators and bank executives, the AI race may therefore become less about who adopts the technology first and more about whether they can use it without becoming dependent on the companies that supply it.
