AI Market Risk Joins CMRA’s Galaxy of Risks™ in Response to Lagarde’s 10/1 Speech
By Peter Niculescu and Leslie Rahl
Two central bank voices on the same day, and the same warning.
On October 1, ECB President Christine Lagarde told the ESRB's annual conference that autonomous AI could amplify market moves, and that AI agents may act in ways the people supervising them never intended and cannot see. She pointed to research in which an AI "trader" acted on an inside tip and then hid its reasoning from its manager.
Galaxy of Risks (click to see some of the history)
(Partial Listing)
The same day, Andreas Viljoen of the Bank of England, writing on the Bank Underground staff blog, described how AI could make the financial system searchable. In one scenario, an agent pieces together scattered weaknesses across banks and funds and finds the path a shock would take. In another, an agent spreads one exposure across derivatives, repo, collateral and affiliated vehicles so that every piece looks modest, every rule is met, and no one sees the whole.
That second scenario will sound familiar to anyone who has had to reconstruct a complex structure after it failed. The questions have not changed since portfolio insurance in 1987. Who understood the strategy? Who was supervising it? What was disclosed? What is new is this: when an algorithm designs the structure, whose intent counts, and can anyone reconstruct what it did?
That is why we have added AI Risk to CMRA's Galaxy of Risks, which we have maintained since the mid-1990s as part of our Risk Advisory practice. It is not a stand-alone risk. It runs through market, model, liquidity, operational, cyber, concentration and legal risk, and it will show up in disputes well before the rules catch up. We point the reader to “Iceberg Risk” which is not new but is relevant.
See the relevant speeches and articles below:
When the financial system becomes searchable – Bank Underground
Frontier AI and the question of governance | Bank of England – the UK's central bank