The Bank of England's Financial Policy Committee has been examining how AI could amplify shocks across UK markets - from herding behaviour in trading algorithms to over-reliance on shared third-party models.

Key takeaways

  • Algorithmic herding is a long-standing concern that AI could amplify (BoE FPC).
  • Model concentration creates a single point of failure if many firms use the same provider.
  • AI-driven trading already dominates US and many European markets.
  • Stress tests assess banks' ability to withstand shocks (annual BoE programme).
  • Investor outcomes can be affected via Liquidity and price-discovery channels.

What is the BoE testing?

Annual stress tests evaluate Capital resilience; thematic exercises now include AI- and cyber-specific scenarios.

Where the risks sit

Shared models, opaque decision-making and rapid algorithmic responses are the main concerns flagged.

What it means for portfolios

Diversified, multi-asset portfolios are typically better placed to ride out short-term liquidity events.

What this means for UK investors

AI stress tests are about systemic resilience, not individual stock picking. For retail investors the take-away is the same: diversify and don't panic-sell into shocks.

Risks to watch

  • Flash-crash style events.
  • Concentration in AI-model vendors.
  • Misalignment of automated risk systems.
  • Liquidity gaps in stressed markets.