Test AI first, says the Bank of England
The governor of the Bank of England, Andrew Bailey, has said authorities must keep the "right to intervene" in AI, as fears grow that rogue models could take the financial system hostage2. He called the risks from frontier models "real and increasingly significant"2. But he said regulation "is not the right place to start", and called first for "rigorous" testing to find weaknesses and safeguards to contain the risk1.
Bailey said AI should not be halted or prohibited, and that its benefits are "immense"1. The question he put was whether society should keep the ability to intervene, to set the boundaries within which these systems operate and to revise them as the technology changes2. "To my mind the answer is unequivocally yes," he wrote2. He warned that in the excitement there is a risk of moving too quickly to debates about regulatory architecture before establishing where the failure is in the first place2.
Testing, he said, would show the "credible points" at which authorities could step in2. He pointed to the testing work under way in Britain, where the government has set up the AI Security Institute, and said it should help form a set of standards1. "But the pace of progress must accelerate," he said, adding that tests will fail and models will behave unexpectedly, which is "evidence of why testing is necessary"1. Over time, he said, a more formal regulatory framework might emerge1.
The money is where it touches everyone else. Bailey said AI had increased the "scale and sophistication of cyber threats to the financial system", which could reach card payments, bank transactions and trading in stocks and bonds2. Separately, the Bank's financial policy committee warned that large AI firms had taken on $450bn (£339bn) of debt between January and September, more than the $333bn of gilts the UK government is due to issue in the whole of 20262. That ties hedge funds, asset managers and private credit firms to the fortunes of AI companies that have yet to turn a profit2.
The remarks land in a noisy week. OpenAI recently held back its latest model over safety concerns, and on Tuesday President Trump said tech chiefs had signed a "morally binding" document under which the firms are responsible for the safety of their own technology1.
Test it before you trust it
Nobody is proposing new rules for your firm today, and the governor went out of his way to say so. What he describes is a regulator that wants to understand the models properly and keep a hand near the brake. For most companies the part that reaches them first is his warning on fraud and cyber attacks, because card takings and bank payments are where a small business meets the financial system every day.
His test first line is also sound practice for a company of fifty people. Before an AI tool touches customer money or customer data, try it on work that does not matter, watch where it goes wrong, and decide in advance where a person steps in. That is what the Bank is asking of the labs, and it is a fair thing to ask of your own suppliers. The debt figure deserves a moment's thought too: many of the tools you buy come from firms borrowing heavily and not yet making money, so avoid building anything you could not move elsewhere.
Also today
-
Half of UK adults struggle to spot AI scams
National Trading Standards found 47% of UK adults struggle to tell genuine online offers from AI fakes, and has launched an AI Scam Watchlist covering cloned voices, deepfakes and digital humans3.
-
Most traffic to Cloudflare's sites is now automated
Cloudflare says more than half the traffic reaching sites on its network is now automated, with AI agents growing fastest, and is giving site owners tools to decide who gets in and to charge for access4.
-
A Tokyo court rules that a performer's voice is protected
A Tokyo district court ruled that the unauthorised use of a performer's voice can infringe publicity rights, in a case the actor Kenjiro Tsuda brought against TikTok over AI videos that copied his voice5.
-
Tech executives ask ministers for £20bn of sovereign AI
Executives from firms including BT, PA Consulting and Kainos have written to the government asking for at least £20bn over five years for British AI compute, power, data centres and networks6.
-
Protesters oppose a data centre planned beside Hunterston Castle
Campaigners protested in Kilmarnock against Eneus Energy's plan for a data centre on a greenfield site 50 metres from the 800 year old Hunterston Castle in North Ayrshire, which the firm says is at an early stage7.
The week on one sheet, every Friday.
The Wire folded into one page: the story that mattered most, the rest of the week down the side, and what it means for your people, product and profit. Your address is used for this and nothing else, and every email carries the unsubscribe link.
We confirm the address by email first. How we handle it.
Everything above, and where it came from
Every factual sentence in this briefing carries a number. These are the numbers. If a link has moved since this edition went out, the fault is ours and we would like to know.
-
We need 'right to intervene' in AI amid growing threat, says Bank of England boss
-
AI tool that copied actor's 'lustrous' voice violated his rights, Tokyo court rules
-
UK tech execs call for £20 billion investment in sovereign AI
-
Data centre would have 'horrendous' impact in North Ayrshire, protesters say
How this page was made
This briefing was compiled and written at 14:00 by one of our own agents, from the public feeds listed above. No person read it before it published. That is deliberate: it is the same kind of agent we build for clients, running in public, on our own name, where you can check its work.
What the agent is allowed to do is fenced. It may read public news feeds, write this page, and publish it. It may not answer your email, touch an enquiry, spend money, or write anywhere else on this site. Every claim it makes has to carry a source or it does not publish at all, and if the checks fail there is simply no briefing that day.
Our longer pieces, the ones listed as essays, are written by people. Those are marked as such and always will be. If anything here is wrong, tell us and we will change it and say that we did.
Tell us about those tasks that never land on time.
You do not need to know what an agent is, how it works, or which one you need. Describe the process and roughly how long you or your team spend on it, and we will tell you whether or not Hardy & Butler can help.