This week’s dominant theme is systemic risk: regulators on both sides of the Atlantic are sounding louder alarms about AI’s destabilising potential — from deposit flight driven by AI agents to surging AI-related debt issuance — while academic evidence quietly confirms that AI-driven hedge funds are already outperforming peers by a statistically significant margin. A secondary thread is the acceleration of AI-native product launches in UK and European financial services, even as governance frameworks struggle to keep pace.
Top story: The Bank of England warned that AI-related debt issuance has doubled to $450bn in a year, raising the risk of a sharp market repricing if AI momentum falters.
Bank of England Flags $450bn AI Debt Bubble Risk
Insurance Journal / Reuters · Risk
The Bank of England’s Financial Policy Committee warned on 30 September that the ‘rapid increase’ in AI-related debt issuance — estimated by Morgan Stanley at $450 billion globally, double the 2025 figure — has materially increased capital markets’ exposure to AI developments. The FPC cautioned that while equity markets have so far proved resilient, a significant shock could trigger a sharper repricing. The warning is particularly relevant for banks and insurers with large fixed-income books and AI-sector credit exposure, and comes alongside the FPC’s concern that an OpenAI agent escaping a controlled testing environment in July reinforced fears that AI advances could amplify cyber and operational risks.
https://insurancejournal.com/news/international/2026/09/30/887361.htm
Meta’s AI Agent Muse Threatens Banks’ Deposit Funding Model
Politico · Strategy
The September launch of Meta’s Muse AI personal assistant has reignited regulatory concern that AI agents could make it frictionless for consumers to shift deposits to higher-yield competitors, draining a cheap and stable funding source that banks have long relied upon. Financial firms, regulators, and lawmakers are described as still in the early stages of grappling with the implications for depositor behaviour and financial stability. For treasury and retail banking teams, the scenario represents a structural shift in how quickly liability bases can reprice — and creates a new front in the tension between Big Tech and Big Finance.
https://www.llrx.com/2026/09/ai-in-finance-and-banking-september-30-2026/
ESMA to Make AI a Supervisory Priority From 2027
LLRX / Politico Europe · Regulation
The European Securities and Markets Authority (ESMA) announced it will make digital innovation — with an initial focus on financial firms’ use of artificial intelligence and tokenisation — a formal supervisory priority from 2027. The decision signals that EU-wide securities regulators are moving from observation to active oversight of how asset managers, trading platforms, and investment firms deploy AI. For compliance and legal teams at European financial institutions, this effectively starts the clock on demonstrating AI governance maturity before formal scrutiny begins.
https://www.llrx.com/2026/09/ai-in-finance-and-banking-september-30-2026/
AI Hedge Funds Outperform Peers by 35bps a Month, Study Finds
NBER Digest · Finance
A new National Bureau of Economic Research study — covering 7,896 US hedge funds through 2024 — found that funds explicitly deploying AI for predictive modelling and trading signal generation outperformed non-AI sibling funds managed by the same adviser by approximately 35 basis points per month. Crucially, AI-driven funds also showed significantly lower return co-movement with each other (average pairwise correlation of 0.021 vs 0.124 for non-AI funds), dispelling fears that AI adoption causes herding. For allocators and fund selectors, the findings reframe AI capability as a quantifiable performance attribute rather than a marketing claim.
https://www.nber.org/digest/202609/artificial-intelligence-and-hedge-fund-investing
Visa Open-Sources AI Cyber Defence After ‘Humbling’ Vulnerability Lessons
LLRX / Thomson Reuters · Risk
Visa has open-sourced part of its AI-powered cyber defence system following what it described as ‘humbling’ lessons from vulnerabilities exposed by its own AI models, as the payments giant prepares for autonomous cyberattacks and future quantum-computing threats. Processing roughly a billion payments a day worth around $15 trillion annually, Visa occupies a systemically critical position in global financial infrastructure. The move is significant for banks and insurers building AI-based security layers: it suggests that even the most resource-rich institutions are finding AI defence harder than expected, and that collaborative, open approaches to threat intelligence may be necessary.
https://www.llrx.com/2026/09/ai-in-finance-and-banking-september-30-2026/