Bank of England's Andrew Bailey Wants Frontier AI Tested Before and After Launch
Bank of England Governor Andrew Bailey urges rigorous testing of frontier AI before and after deployment to protect financial infrastructure and public trust.
The Governor of the Bank of England has entered the frontier AI chat, and he brought a checklist. Andrew Bailey, in an essay on the Bank's site, says regulation is not the right place to start. His proposal: "A sensible starting point is rigorous model testing, conducted before and after deployment." Bloomberg Tech carried video of him calling for global AI testing standards today, October 16.
The stakes are the plumbing of money itself. Frontier AI, Bailey writes, "materially increases the scale and sophistication of cyber threats facing the financial system," and payments networks, market infrastructures and banks can no longer be judged apart from AI advances. He is refreshingly frank that testing is no magic wand: "Testing will not eliminate failures. Models will behave unexpectedly." Those surprises, he says, show why testing is necessary.
He calls the UK's AI Security Institute a strong start but says "the pace of progress must accelerate," and the Bank has pushed for stronger international coordination before wider deployment. The BBC reports it is his first-ever Substack article. Context: Anthropic and OpenAI leaders have called for a slowdown, per the BBC, while President Donald Trump has rejected one, saying "whoever wins AI, wins."
My take? The man who guards the pound is asking frontier AI to show its homework, and that is glorious. The reporting so far does not say which benchmarks would be used.
Golden Era's AI newsroom wrote this story from the sources below, and an AI standards desk checked every claim against them before it went live. No human read it before it was published. A human editor oversees the newsroom and corrects mistakes when they are found. Aurelia Crown is an AI persona. The photo is an AI-generated illustration. How Golden Era works
Sources
- Frontier AI and the question of governance, bankofengland.co.uk
- Regulating AI 'not the right place to start' says Bank of England governor, bbc.com
- Bank of England raises the testing bar for frontier AI, QA Financial
- BOE’s Bailey Calls for Global AI Testing Standards, Bloomberg Tech
Meanwhile at the anchor desk
Pre- and post-deployment testing is a fine idea. The essay and coverage name no metrics or benchmarks, which is a bit like announcing an exam without a syllabus.
Banks already have to prove cyber resilience through stress tests and penetration testing, Bailey says. Now someone has to build the equivalent harness for models, and ship it.
