Bailey warns AI valuations face four financial stability risks
BOE Speech

Bailey warns AI valuations face four financial stability risks

Bank of England Governor Andrew Bailey warned in Zurich that AI valuations face risks from overstated returns, delayed gains, misplaced confidence in winners and lost public trust. He also defended central bank independence and post-crisis regulation.

Independence rests on public legitimacy

Bailey said conflicts in Europe and the Middle East create supply shocks that raise inflation and weaken activity, complicating monetary policy.

AI investment, particularly in business services, has meanwhile supported stronger-than-anticipated UK activity.

He argued that difficult conditions strengthen the need for robust independent institutions, accountable to the whole population.

Monetary and financial stability both protect the value of money, but financial stability decisions attract more lobbying and political contestation than rate decisions.

As memories of the financial crisis fade, pressure to ease regulation is increasing.

Bailey credited post-crisis reforms with financial resilience and said sustained growth depends on stability.

Four routes to valuation losses

Bailey identified four risks around a possible AI asset-price bubble, while stressing that identifying bubbles is imprecise.

Markets might overvalue uncertain returns; gains might arrive later than investors anticipate; valuations might assume every company wins; or lost public trust might undermine prices.

He cited Edison and electricity to illustrate delayed returns and Netscape to illustrate competition.

AI, digital technology and robotics offer productivity gains, he said, but frontier AI also increases cyberattack risks.

Trust needs human control

Bailey argued that AI's commercial success depends on public trust, linking technology policy to financial stability.

He called for more model testing before and after release, better incident reporting, monitoring and follow-up.

In his account, assured human intervention underpins trust in AI as a tool for people.

Report an error