JP Morgan Chase most advanced in AI as UK banks group in chasing pack

You have probably watched your organisation announce an AI strategy, appoint a chief AI officer and roll out a pilot or two, only to wonder whether anyone else is moving faster. Evident's 2026 AI index offers an answer: most are not moving fast enough, and a handful are pulling away. JP Morgan Chase retained the top position for artificial intelligence capabilities among 50 global banks, according to the benchmark. Six US-headquartered institutions and two Canadian banks filled out the top ten, alongside Switzerland's UBS and Australia's CommBank. The UK's largest banks (HSBC at 11, Lloyds Banking Group at 15, NatWest at 17 and Barclays at 19) clustered in the chasing pack, competent but not leading.

The gap matters because AI capabilities advanced nearly three times faster over the past year than the average across the previous three years, according to Evident. Alexandra Mousavizadeh, co-CEO at Evident, said that "banks across the AI index are moving faster than at any point since we began measuring." Speed compounds. An institution that lags today will find the distance harder to close tomorrow, not because the technology is inaccessible but because the leaders are building organisational muscle (data pipelines, risk frameworks, talent density) that cannot be bought off a shelf.

What separates the leaders

The benchmark reveals two characteristics that distinguish the top cohort. First, they are reporting or projecting returns. Twelve of the 50 banks now claim measurable value from AI activities, up from eight the previous year. Lloyds Banking Group joined that group, stating that all customer interactions will be supported by AI by 2030 and targeting £100 million in value generation. That figure is not a moonshot; it is a budget line tied to specific use cases. The discipline required to name a number and defend it internally forces clarity about where AI will work and where it will not.

Second, the leaders invest hardest in guardrails. Evident found that 80 per cent of the top-tier banks use sophisticated controls (checks on training data, output validation and live monitoring) compared to 40 per cent of the rest. Responsible AI principles are now near-universal, adopted by 49 of the 50 banks surveyed, an increase of 16 since 2023. But principles are not controls. A policy document that commits to fairness and transparency does little if the institution lacks the technical infrastructure to audit a model's behaviour in production or the governance process to pull it offline when something breaks.

The gap between principle and practice explains why public debate about AI safety often feels disconnected from what enterprises actually do. Regulatory scrutiny is mounting, and banks that treat compliance as a checkbox exercise will find themselves outpaced by competitors who embed risk management into the development cycle. The institutions pulling ahead are not waiting for regulators to tell them what to monitor; they are building the monitoring capability now, because they know that a model that drifts or hallucinates in a customer-facing application will cost more than the engineering time required to prevent it.

What this means for L&D and capability leaders

The Evident benchmark is a proxy for organisational readiness, not just technology adoption. The banks that rank highest are not necessarily running the most advanced algorithms; they are the ones that have trained enough people to use AI tools safely, built enough process to govern their deployment and embedded enough measurement to know whether the investment is working. That readiness is a capability problem, and capability problems belong to learning and development, innovation and transformation leaders as much as they belong to chief technology officers.

If your organisation is in the chasing pack, the question is not whether to adopt AI (that decision has been made) but whether you are building the infrastructure to adopt it well. That infrastructure includes technical guardrails, but it also includes the human systems that make those guardrails effective: the training that helps a product manager understand when a model is appropriate for a task, the governance that ensures a risk officer can challenge a deployment decision and the measurement discipline that turns a pilot into a scalable programme.

The UK banks in the middle of the Evident ranking are not failing; they are competent. But competence is not enough when the leaders are accelerating. The institutions that will close the gap are the ones that treat AI capability as a strategic priority, not a technology project. That means investing in the boring work of data quality, process design and skills development, and it means doing that work faster than the competition. The alternative is to watch the distance grow.

Sources:

• Computer Weekly, 6 October 2026