HM Treasury’s two AI Champions have published a plan setting out how the government, regulators and financial firms should speed up the safe adoption of artificial intelligence, with 10 recommendations the government says it has accepted where they fall to it.

The Financial Services AI Adoption Plan, published on 14 July, was written by Harriet Rees, group chief information officer at Starling, and Dr Rohit Dhawan, head of AI and advanced analytics at Lloyds Banking Group. Both were appointed to the champion roles by the Treasury in January 2026.

Key takeaways 

  • HM Treasury’s AI Champions, from Starling and Lloyds, set out 10 recommendations to speed AI adoption; the government accepted those addressed to it.
  • The plan urges faster rollout of the Critical Third Party regime, a voluntary industry AI assurance scheme with incident-sharing, and a trust framework for agentic payments.
  • A Bloomberg survey of 500 London finance staff found 88% use AI weekly, but 73% say it hasn’t cut their workload; their top worry is AI-enabled crime, not job losses.

 

Chancellor Rachel Reeves said the plan was “central” to establishing the UK “as the fastest adopter of AI in the G7”, part of what she framed as a wider push for domestic AI sovereignty. 

What the plan recommends

The 10 recommendations run across regulation, resilience, skills and payments, and aim to accelerate AI adoption across UK financial services. They ask regulators to set clearer expectations and more accessible support for firms building AI, and call on the Financial Conduct Authority to review how large language models generating financial guidance sit against the regulatory perimeter and consumer protection. 

On resilience, the champions want the government to accelerate the Critical Third Party regime for the AI and cloud providers the sector increasingly depends on, and to stand up a voluntary industry-led scheme for assuring third-party AI models, alongside a repository for sharing AI incidents and near-misses across firms.

The plan also proposes a sector-wide AI skills plan, wider take-up of the Financial Services Skills Commission‘s skills compact and visa adjustments to attract specialists, and points to a Treasury consultation to build a “trust framework” for agentic payments, where AI agents transact on a customer’s behalf.

Rees said the message from industry had been consistent. “Financial services firms are motivated to adopt AI at scale and to take a world-leading position,” she said. “There’s broad agreement that we have the regulatory foundations to do this safely and that we now need to move forward with greater clarity, focus and ambition.”

A workforce already ahead

The plan lands as separate research suggests London’s finance workforce is already well ahead of it. A Bloomberg survey of 500 London-based financial services professionals, conducted by Stack Data Strategy between 16 April and 25 May 2026, found that 88% already use AI tools daily or weekly, and not one respondent said they had yet to start learning to use them.

AI fluency is fast becoming a baseline expectation. The survey found 85% believe AI skills will soon outweigh traditional ones for their next promotion, and 69% think colleagues who resist building those skills have no place at their firm.

The gains are uneven. While 90% said AI had improved their productivity, nearly three-quarters (73%) said it hadn’t reduced their workload. Of those who had freed up time, 51% were redirecting it towards higher-value work such as strategy, decision-making and client relationships.

Job displacement barely registered. The loss of entry-level and graduate roles was among the least-cited worries, selected by 30% of professionals, against 45% who named AI-enabled cyberattacks and financial crime as a top concern, the same resilience risk the plan’s assurance and incident-sharing proposals target.

Amanda Stent, head of AI strategy and research in the CTO office at Bloomberg, said the findings should shape how policy is written. “A model may detect a pattern; it takes a trader or analyst with years of experience to understand whether that pattern is a meaningful signal or just noise,” she said. “Any government plan for AI adoption in financial services needs to account for that.”