Bank of England Deputy Governor: AI Autonomous Trading Threatens Financial Stability, Circuit Breaker Mechanisms Under Study
nashnova research
BoE Deputy Governor Sarah Breeden warned that AI agents trading autonomously could amplify volatility and trigger market turmoil — the regulatory window is narrowing, and the central bank is exploring a kill-switch mechanism that remains undefined.
What exactly is she worried about?
Breeden's core concern is a machine-speed herd effect: multiple AI agents read the same market signal, make near-identical trades, and create a feedback loop — signal → trade → price move → more trades — that amplifies asset-price swings.
The deeper risk is emergent coordination among AI systems — not deliberate collusion, but algorithmic logic converging into collective action that destabilises markets further.
In plain terms = one trader following the crowd is manageable; thousands of machines doing it in milliseconds, with no one able to call a halt, is not.
What does the "kill switch" mean?
Breeden backs a circuit-breaker mechanism: when AI-driven trading creates market stress, regulators could restrict or suspend those trades.
This means → regulators want an emergency brake that intervenes before AI trading spirals — similar to existing stock-market trading halts, but designed specifically for AI agents.
Yet Breeden herself conceded: "I don't yet fully understand how it would work in practice." The direction is set; the details are blank.
Can humans actually oversee AI in real time?
Breeden acknowledged an inherent limitation: humans cannot fully supervise AI agents in real time — trading speed far exceeds human reaction time.
This means → the "fix it after it breaks" approach will not work; the institutional framework must be in place before AI goes live, making pre-emptive design far more important than post-crisis repair.
This reflects a deeper tension: regulation moves on an annual cycle, AI capability iterates monthly — the time gap itself is a risk.
Can international regulation keep up?
The Financial Stability Board published a report in June proposing a global framework for responsible AI adoption by financial institutions.
But Breeden's remarks signal that the pace of technological change already outstrips regulators' ability to respond, and the international community has yet to converge on a control path.
In plain terms = everyone agrees AI finance needs rules, but *what* rules, *how strict*, and *who enforces* remain unsettled — and AI autonomous trading will not wait for consensus.
市场有风险，内容仅供研究参考，不构成投资建议。