Banks Double Down on AI as Fintechs Surge and Regulators Tighten Oversight

Global banks are stepping up their commitment to AI as a strategic priority. In one of the clearest signs of this trend, HSBC announced plans to launch a Global AI Centre of Excellence in Singapore, hiring over 100 specialists in areas like data science, AI governance, and human-centric design ([1]). The new hub – led by HSBC’s first-ever Chief AI Officer, appointed in March ([2]) – will focus on initiatives such as AI-enhanced wealth management consultations, “agentic” treasury solutions, and AI-enabled digital payments innovations ([3]). HSBC’s leadership chose Singapore for its supportive ecosystem and talent pool, and the bank has already struck partnerships with firms like Google’s DeepMind and French AI startup Mistral to accelerate development of scalable AI solutions across its global operations ([4]). As HSBC’s Group CEO emphasized, the goal is to deploy AI "safely, in real time and at scale, while keeping human judgement, decision-making and accountability at the core" ([5]).

The scale of AI adoption cited by other major banks underscores that this technology is moving from pilot projects to an integral part of daily operations ([6]). Bank of America’s CEO Brian Moynihan told investors that over 200,000 of the bank’s employees now use AI-powered tools, together generating more than 400,000 prompts to its internal AI systems each day ([7]). The bank has more than 300 AI use cases approved (including 114 using generative AI), with 34 already fully implemented in production processes across the organization ([8]). JPMorgan’s Jamie Dimon said his firm has nearly 1,000 AI applications live, embedded in everything from risk modeling and fraud detection to marketing and document analysis ([9]). Meanwhile, Wells Fargo recently rolled out an "AI-powered Teammate" virtual assistant for its staff, which CEO Charlie Scharf noted is improving employee productivity across the company ([10]). And at Citigroup, CEO Jane Fraser highlighted that almost 90% of employees are using the bank’s AI tools – boosting productivity, speeding up product delivery, and enhancing client experiences ([11]). Collectively, these disclosures from banking leaders signal a watershed moment: AI is now being treated as a core infrastructure and operational backbone of modern banking, rather than a niche experiment ([12]).

This deep integration of AI is already prompting shifts in bank workforce and leadership models. Institutions are creating new C-suite roles to drive AI strategy, exemplified by HSBC’s appointment of a dedicated Chief AI Officer and Bank of Ireland hiring its first Chief AI Officer in July ([13]). Even outside of banking, major asset managers are realigning their leadership to emphasize technology’s new centrality; for instance, UK investment firm M&G just named a Chief Technology and AI Officer to oversee its digital transformation ([14]). These moves reflect an industry-wide recognition that advanced AI capabilities – from machine learning in risk management to generative AI in customer service – are now critical to competitive advantage and require executive oversight.

Yet even as they scale up AI, bank executives are mindful of managing its costs and risks. Deploying large language models can be expensive, and the industry is learning that more AI is not always better. Dimon cautioned that AI, while promising, remains costly and may not boost bank profit margins in the near term ([15]) – he predicts the primary benefits will accrue to customers through improved services, rather than immediate cost savings for the bank. Indeed, after a period of enthusiasm where some firms deployed AI tools everywhere possible – a trend dubbed “tokenmaxxing” in reference to maximizing AI usage – many companies are now pulling back due to skyrocketing cloud bills and only modest productivity gains ([16]). This more measured approach suggests that success will come from targeted, efficient AI applications that deliver clear ROI, rather than indiscriminate experimentation. Forward-looking banks are thus doubling down on AI with a balance of ambition and pragmatism: invest in bold innovation, but track the business value closely and ensure strong controls.

Even as banks invest heavily in AI, a new generation of fintech and insurtech firms is leveraging AI to challenge the industry status quo. The past two days saw a remarkable example: Corgi, a little-known AI-powered insurance startup, has reportedly achieved a $4 billion valuation after raising capital four times in just a few months ([1]). According to reports, Corgi’s investors have poured in successive funding rounds – some spaced mere weeks apart – roughly doubling the company’s valuation each time ([2]). The full-stack insurer uses artificial intelligence to automate underwriting and claims, promising to deliver faster quotes, more flexible policies, and speedy payouts for businesses in sectors from trucking to small business coverage ([3]). The enthusiasm around Corgi is fueled by its dizzying growth: the company’s annual revenue run-rate, $40 million at the start of the year, is now on track to reach $450 million by the end of 2026 ([4]). This kind of 10× jump in revenue in under 12 months is virtually unheard of in the insurance industry, underlining how a nimble AI-driven entrant can quickly scale and attract capital by addressing pain points that incumbents have struggled with.

In the banking sector, another fintech is crossing a different kind of milestone. US-based online lender Upstart, known for its AI-driven credit decisioning models, just received a conditional green light from the Office of the Comptroller of the Currency to launch its own nationally chartered bank ([5]). If the remaining approvals are secured, “Upstart Bank, N.A.” would become the first full-service national bank built from scratch around AI-powered underwriting and risk models ([6]). By obtaining a bank charter, Upstart aims to cut operational costs and expand its lending nationwide, using AI to price credit more efficiently and potentially undercut traditional lenders. The move demonstrates regulators’ growing comfort with novel, AI-centric business models in financial services – and portends stiffer competition for established banks, especially in consumer and small-business lending.

Meanwhile, digital-first banks are developing advanced AI capabilities in-house that rival or exceed those of larger institutions. UK-based fintech giant Revolut recently announced a proprietary “AI brain” called PRAGMA – a foundational model trained on all of its customers’ transactional, behavioral, and support data ([7]). This unified AI system, built in partnership with NVIDIA, has already led to a 64.7% improvement in fraud detection rates and a 16% better accuracy in credit risk prediction, according to a report from Forbes ([8]). By vertically integrating AI development and tailoring models to its own rich data from 70 million users, Revolut has set a new benchmark for what a digital-era financial institution can achieve with AI innovation ([9]). Incumbent banks that rely mostly on third-party AI tools may find themselves at a disadvantage if fintechs can offer smarter, more personalized services powered by these homegrown AI models.

Collectively, these fintech advances are a wake-up call for the industry. Established players face intensifying pressure to match the speed and creativity of AI-enabled challengers. Some banks are responding by partnering with or investing in AI startups, while others are accelerating their own innovation programs. But the message is clear: AI-driven disruption in financial services is no longer on the horizon – it’s already here, reshaping the competitive landscape in real time.

As AI becomes ubiquitous in finance, regulators worldwide are racing to ensure the technology is deployed responsibly. In a notable development, the Reserve Bank of India (RBI) has proposed a comprehensive new framework for model risk management that explicitly covers AI and machine learning models in banking ([1]). The draft guidance calls for measures such as firm-wide inventories of all models in use, risk-based tiering of models, and board-approved governance policies. It would hold banks accountable for outcomes from third-party or outsourced AI models (mandating independent validation regardless of vendor assurances) and impose strict requirements for high-risk AI models – including mandatory explainability, fairness and bias testing, controls on “drift” and hallucinations, robust human oversight, and the ability for customers to have AI-driven decisions reviewed by a human ([2]). Once finalized, these rules would set a high bar for AI governance, likely influencing regulators in other jurisdictions.

Europe is also on the cusp of major AI regulation affecting finance. The European Union’s AI Act, slated for phased implementation, will begin enforcing new compliance obligations for providers and users of high-risk AI systems on August 2, 2026 ([3]). This includes many AI applications common in financial services – from credit scoring and fraud monitoring to algorithmic trading and insurance risk models – which are classified as “high risk” under the EU rules. In anticipation, European financial firms have been preparing to meet requirements around transparency, data quality, and human oversight for these AI tools. Notably, recent amendments to the Act have provided clarity for insurers and delayed some deadlines for certain provisions, giving institutions more time to adapt their models and address concerns about bias and accuracy ([4]).

Across the Channel, the UK’s Financial Conduct Authority (FCA) is experimenting with a more collaborative approach to AI oversight. It recently expanded its “digital sandbox” programme – dubbed the Supercharged Sandbox – to include cutting-edge AI vendors. For example, California-based AI firm Anthropic has been brought in to offer its large language model platform (Claude) for participants in the sandbox, allowing banks and startups to test AI-driven financial products under the regulator’s supervision ([5]). By directly engaging with AI developers and financial firms in a controlled environment, the FCA aims to foster innovation while gathering insights that could shape future regulatory frameworks. The UK is also examining broader implications of AI: the newly published Mills Report commissioned by the FCA outlines a 2030 vision where AI is deeply embedded in retail financial services and provides recommendations on how regulators should respond ([6]).

Meanwhile, U.S. regulators have yet to issue similarly prescriptive AI rules for banks and investment firms, but they are not standing idle. The Securities and Exchange Commission, for instance, has not introduced any AI-specific disclosure requirements as of mid-2026, instead applying existing standards (such as anti-fraud and consumer protection laws) to AI deployments ([7]). Federal banking agencies have signaled that guidance on AI and model risk management is forthcoming, emphasizing that banks must ensure their AI algorithms do not violate fairness and transparency standards. The overall regulatory tone in the U.S. is one of cautious observation coupled with case-by-case enforcement, exemplified by the specter of legal action over AI bias in consumer lending and hiring practices.

In addition to formal regulations, authorities are highlighting emerging AI-driven threats that financial institutions need to manage. Hong Kong’s Securities and Futures Commission (SFC), for example, warned in a June circular that “frontier” AI models are dramatically lowering the cost and skill threshold for cybercriminals, enabling more frequent and sophisticated attacks ([8]). These include automated hacking, phishing campaigns, and even deepfake voice and video impersonation used to deceive bank customers or trigger fraudulent transactions. The SFC urged financial firms to bolster patch management, access controls, third-party risk oversight, and incident response plans to handle AI-enabled threats ([9]). The rise of deepfake fraud is a particular concern globally, with Deloitte projecting the deepfake detection market to reach $15.7 billion by 2026 ([10]) as institutions invest in technologies to verify identities and protect customers. Whether addressing sophisticated cyber risks or potential algorithmic bias, the clear signal from regulators is that AI in finance must evolve within a strong governance framework. Senior executives should expect more scrutiny of how their organizations deploy AI, and ensure they can explain and justify algorithmic decisions to both supervisors and customers.