Stock Markets July 14, 2026 01:45 PM

UK banks urged to build domestic AI capabilities after limited Anthropic access

Government-appointed banking adviser warns Britain must develop models, infrastructure and skills to reduce reliance on U.S. providers

By Jordan Park
Share
Twitter Reddit Facebook LinkedIn
JPM

A government-appointed banking adviser has highlighted gaps in the UK’s artificial intelligence capabilities after only a small number of British banks gained access to Anthropic’s new model, Mythos. Harriet Rees, chief information officer at Starling Bank and the finance ministry’s AI champion, said the UK needs to create its own AI infrastructure and talent quickly to avoid dependence on overseas technology providers. Mythos, released in April to a limited group that included JPMorgan, is viewed as a leading tool for identifying cybersecurity weaknesses and enhancing banks’ security responses.

UK banks urged to build domestic AI capabilities after limited Anthropic access
JPM
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • Access to Anthropic’s Mythos was limited, with the April release including U.S. lender JPMorgan among the select users.
  • Harriet Rees, Starling Bank CIO and the finance ministry’s AI champion, urged the UK to build domestic AI infrastructure, models and skills to reduce reliance on U.S. technology.
  • The restricted distribution of Mythos affects banks’ cybersecurity capabilities, as the model is seen as a leading tool for identifying vulnerabilities and accelerating defensive responses.

UK banks’ limited access to Anthropic’s new AI model, Mythos, has raised concerns about Britain’s readiness to rely on overseas technology for critical cyber-defence tools. Harriet Rees, chief information officer at Starling Bank and the finance ministry’s designated AI champion since January, said the situation underlines the need for Britain to invest in its own AI infrastructure, models and skillsets to lessen reliance on U.S. technology suppliers.

Mythos, which was released in April to a narrowly defined group of users that included American lender JPMorgan (NYSE:JPM), is regarded within the industry as a particularly advanced instrument for spotting cybersecurity vulnerabilities. Proponents say that access to the model helps banks identify weaknesses more quickly and strengthens their overall security posture.

Rees said that only a small number of UK banks received access to Mythos, and those that did were mainly the British operations of U.S. lenders. She argued that this access gap points to a broader strategic challenge for the UK financial sector and the nation’s technology ecosystem.

"Now is the time for us to think very strategically about what we need to do to protect our leading position moving forward. Time really is of the essence, we don’t have two years here," Rees said.

Her comments frame the issue as both an infrastructure and skills problem: to avoid an overreliance on foreign AI offerings, the UK would need to construct its own models and the systems to host them, while also ensuring that the workforce possesses the capabilities to deploy and manage those systems effectively. Rees’ role as the finance ministry’s AI champion places her in a position to advise on the scale and speed of such efforts.

The practical impact of restricted access to an advanced model like Mythos is concentrated on banks' cybersecurity programmes. With fewer domestic institutions able to deploy the tool, those banks could face slower identification of vulnerabilities compared with peers that obtained access. At the same time, the situation highlights the intersection of national technology strategy and the operational needs of the financial sector.

Rees’ remarks prompt a policy question about whether developing local AI infrastructure and models should be prioritised to shore up the UK’s financial system and reduce dependence on external technology vendors. She emphasised urgency in planning and execution, noting a limited time window to act.


Key facts in this report: Mythos was released in April to a limited circle that included JPMorgan. Only a small number of UK banks received access, mainly UK arms of U.S. lenders. Harriet Rees, Starling Bank CIO and the finance ministry’s AI champion appointed in January, called for Britain to build AI infrastructure, models and skills to reduce dependence on U.S. providers.

Risks

  • Continued dependence on foreign AI providers may leave UK banks less able to access cutting-edge cybersecurity tools quickly - this primarily affects the banking and cybersecurity sectors.
  • A shortfall in domestic AI infrastructure and talent could slow the UK’s ability to deploy its own models and systems, impacting the financial sector’s operational resilience and technology markets.
  • If only a subset of institutions retain access to advanced tools like Mythos, competitive imbalances could emerge within the banking sector, particularly between UK operations of U.S. lenders and domestic banks.

More from Stock Markets

Goldman Sachs Shares Jump After Record Quarterly Results and Dividend Hike Jul 14, 2026 Yum! Brands Shares Slip After Taco Bell Pulls Items Amid Cyclospora Investigation Jul 14, 2026 CVS Caremark Agrees to FTC Terms on Rebates and TrumpRx Payments Jul 14, 2026 CVS Caremark Agrees to FTC Settlement, Will Count TrumpRx Purchases Toward Deductibles Jul 14, 2026 Medical device shares slide after HCA flags weaker surgical activity Jul 14, 2026