UK banks are already reviewing their appetite for AI-linked lending and investment, according to the head of the Prudential Regulation Authority. Guidance on frontier AI and quantum risks from the PRA and FCA is due later this month, putting a practical question on banks’ desks: which clients and dependencies are worth the exposure?
Watch Desk analysis
What happened
Katharine Braddick, who took over as PRA chief in July, told Bloomberg that firms are considering which AI-related clients to support. The Bank of England’s review has not found that lenders would amplify an AI shock, Resultsense reports, but Braddick said scrutiny of firms’ risk appetite is already feeding through.
The PRA and its conduct-regulator counterpart, the FCA, plan to issue guidance on frontier AI and quantum computing risks later in November. Braddick described it as “quite concrete” and said she hoped it would begin an iterative exercise with the industry. The guidance is expected to set governance expectations; it is not a new capital requirement, and the Bank has no current plan to use capital rules to force lenders to cut AI exposure.
Why it matters
AI risk is becoming a banking question before it becomes a reason for regulators to impose a broad lending brake. A bank can be exposed through loans to AI companies, reliance on AI suppliers, or the operational and cyber risks those systems bring. The upcoming guidance could give boards and supervisors a clearer basis for asking how those risks are assessed.
There is a distinction worth keeping: firms reviewing their risk appetite is not evidence of a system-wide retreat, and a review that found no likely amplification is not a guarantee against future shocks. The practical signal is that lenders are beginning to make choices client by client.
Our read
This is a more useful story than the latest round of “AI bubble” shouting because it points to action inside banks, while stopping short of claiming a regulatory crackdown. The guidance should make those judgements easier to scrutinise. For now, banks with large AI-linked loan books or heavy reliance on AI suppliers should expect questions about both credit and cyber resilience.
What to watch
- What governance expectations the PRA and FCA set in their guidance later this month.
- Whether the regulators ask banks to map exposures to AI clients and suppliers in more detail.
- Whether firms’ changing risk appetite shows up in lending decisions, rather than remaining a boardroom discussion.
Discussion spark: Should banks treat AI exposure mainly as a credit risk to manage client by client, or as a shared systemic risk that warrants common limits?
Sources and evidence
- PRA chief says UK banks are trimming their AI exposure – Resultsense (13 November 2026, 12:06 UTC)
- Royal Navy signs AI software deal with Arondite for UK Commando Force – Resultsense (13 November 2026, 12:06 UTC)
- NHS England removes list of 137 trusts it said benefit from Palantir – Resultsense (13 November 2026, 12:06 UTC)
- Google launches Gemini 4 Hydrogen, a fast model for agent workloads – Resultsense (13 November 2026, 12:06 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.