The UK's Financial Conduct Authority is preparing for a financial services landscape where AI does more than answer questions or support customer service. Increasingly, AI systems may search for financial products, manage payments, support investment decisions, and execute transactions on a consumer's behalf. That shift opens the door to real innovation, but it also raises hard questions about responsibility and consumer protection.

In a speech delivered on October 19, 2026, at the Bloomberg UK Policy Series, Jessica Rusu, the FCA's chief data, information, and intelligence officer, set out the regulator's priorities for safe AI adoption, focusing on consumer outcomes, governance, and operational resilience.

One finding stands out: four in five less-experienced investors aged 18 to 40 have used AI to help with investing. Yet only 40% of consumers correctly recognize they may have no recourse if something goes wrong after acting on advice from a general-purpose AI tool, a real gap between adoption and understanding.

The FCA is not trying to stop AI use in finance. It wants firms and technology providers to build systems that support innovation while protecting consumers and preserving accountability. As AI agents start acting on behalf of users, firms will also need to know what a system is authorized to do, how its actions are monitored, and what happens when something goes wrong.

The Consumer Protection Gap in AI Investment Advice

General-purpose AI tools have become an easy way to access financial information. Consumers use conversational systems to explore investment concepts, compare products, and ask questions they once would have taken to a professional. That access can genuinely help people engage with investment decisions. But the FCA is concerned about the line between general information and regulated financial advice, since a consumer may not always know whether an AI-generated response counts as education, a personalized recommendation, or something within a regulated activity. That distinction can shape what protections apply if the advice leads to a loss.

The FCA's answer is not to expand its regulatory reach but to sharpen the existing boundary. It is calling on AI providers to add clear, proportionate, and consistent disclaimers explaining the limits of the financial information their models generate. Such disclosures could help, though they alone cannot guarantee understanding, and a polished AI response can sound authoritative even when the model lacks key context.

For fintechs building AI into consumer products, the user experience must not obscure what kind of service is being offered. Firms need to think through how users interpret recommendations and when a conversation should escalate to a human. The FCA already runs initiatives such as InvestSmart and its Firm Checker tool, and Rusu said the regulator plans to build on that work rather than create a new public tool, including through IOSCO's first TechSprint and its AI Lab, which helps investors spot scams and challenge unreliable AI outputs.

FCA Says 86% of Financial Firms Use AI.

Consumer-facing AI is only part of the picture. Rusu referred to a joint AI survey with the Bank of England, due for publication the following week, which found that 86% of firms already use AI. Since the speech came before that publication, the figure should be attributed to Rusu's remarks rather than treated as an independently confirmed result here.

That adoption level illustrates the FCA's central concern: AI is no longer something firms can watch from the sidelines. It supports fraud detection, customer service, document processing, and compliance monitoring, and well-implemented systems can lift productivity. But deployment also brings responsibility, since firms must understand the risks they are taking on and keep people accountable for outcomes.

Three Priorities for Responsible AI Deployment

Rusu's speech identified three areas of good and poor practice drawn from the FCA's observations of firms testing. AI: consumer centricity, accountability and governance, and operational resilience.

On consumers, the FCA wants AI to improve outcomes, not just automate existing processes. It has seen firms use AI for inclusive product design, including one chatbot that used a live list of vulnerability signals to flag consumers who might need a human adviser. But it also flagged poor practices: building tools without a clear need, excluding vulnerable consumers by default, overstating what systems can do, and launching automated journeys without enough testing.

On governance, the FCA sees under-investment across the industry. AI systems touch decisions across many teams, and without clear ownership, it becomes hard to know who is responsible when something fails. Rusu said firms need clear accountability, risk-based controls, and escalation triggers, with senior managers staying personally accountable. One firm set up an independent ethics panel to review AI use cases before deployment; others have tried to retrofit governance afterward, letting oversight lag behind the system it is meant to control.

On resilience, the FCA wants firms to stress-test AI systems, plan for failure, maintain fallback options, and understand their dependence on suppliers, including reliance on a single model, alongside cyber insurance. A system that performs well in testing may act differently once it meets new data, so firms should plan how to keep operating if a tool becomes unavailable or unreliable.

Agentic Finance Brings a New Set of Risks

The FCA's agenda goes beyond AI systems that inform or support human decisions. Rusu described a shift toward agentic finance, where systems increasingly take action on a user's behalf: searching financial products, managing payments and investments, and interacting directly with other agents. An agent might help a consumer compare products or carry out an authorized task within set parameters; in business settings, agents could support transaction monitoring or operational workflows.

Greater autonomy changes the risk profile. When an AI system only provides information, a person can still review it before acting. When an agent can execute a transaction, a single error can have direct consequences. A firm must know which agent is acting, who authorized it, what permissions it holds, and how its actions can be traced, and it must respond quickly if an agent exceeds its scope.

The FCA says it will work with industry, other regulators, and frontier AI laboratories to build a trusted framework for agentic finance, covering agent identity, delegated authority, and accountability, and building consumer protection into agentic systems from the start. This matters for fintech because the next generation of financial services may run through software agents operating between customers and institutions.

AI Failures and Near-Misses Become a Regulatory Priority

The shift toward autonomous systems also changes how firms should handle incident reporting. Rusu said poorly governed AI could create consumer harm and operational failures faster and at greater scale. Existing rules already require firms to report major incidents, and the speech argues this should extend to significant AI failures and near-misses, which matter because a system can misbehave without causing immediate loss or visible disruption. An AI agent might attempt an action outside its permissions or produce a wrong recommendation a human catches before execution; documenting such cases strengthens controls. The FCA called near-miss reporting for agentic AI a growing national security priority, though the speech sets out no standalone reporting rule, signaling instead that firms should take AI failures seriously within existing responsibilities.

The FCA Is Building Its Own AI-Enabled Supervision

The regulator is applying AI to its own work too. Rusu described the FCA's ambition to become a smarter regulator, using technology to gather evidence, connect intelligence, and spot early warning signs, aiming to speed up oversight rather than replace its own judgment.

Two examples show the scale of the work. The FCA is building AI tools to oversee roughly 60,000 anti-money laundering firms expected to come under its remit in 2028, and it is deploying automation and agentic tools, including a market first-responder agent, to help supervise around 9,000 small wholesale-market transaction-reporting firms, freeing supervisors for more complex work. Success will depend on data quality and proper validation. The regulator is also investing in machine-readable regulation, including the FCA Handbook API launched in August, with data-collection APIs in development.

The speech also touched on frontier AI and cybersecurity. Rusu said the FCA is working directly with banks to understand how frontier models are changing cyber defense, helping firms identify vulnerabilities and recover faster. Advanced AI tools speed up remediation, but more capable models also introduce new security risks, especially when connected to sensitive systems. The FCA's approach is to engage directly with firms rather than rely on broad claims.

What This Means for Fintech Companies

The FCA's priorities touch firms across payments, lending, wealth management, banking, and financial infrastructure. Companies building AI-enabled products need to treat consumer protection as part of product design, explaining what a system does and offering clear routes to human help. Companies using AI internally still need clear ownership, testing before deployment, and fallback arrangements for critical services.

For businesses building agentic products, identity and delegated authority matter most, since any system that can initiate a financial action needs clear permissions and a way to establish who authorized it. Firms should also assess their dependence on third-party models and suppliers, since a change or outage there can ripple through services built on top of it. Not every application faces identical requirements, but the direction is clear: AI should be deployed with accountability and resilience built in from the start.

Conclusion: AI Innovation Must Keep Pace With Accountability

The FCA's latest position reflects a financial sector moving past AI experimentation. Consumers are using general-purpose tools to inform investment decisions, firms are weaving AI into their operations, and autonomous agents are starting to reshape financial tasks. The regulator sees real potential here, including better productivity and more effective supervision, but it also warns that innovation without proper controls can expose consumers and firms to new risks.

The figures Rusu cited, four in five younger investors using AI alongside only 40% who understand the limits of recourse, show exactly why consumer understanding matters right now. The FCA's response pairs clearer disclosures with practical examples of good and poor practice, stronger governance, operational resilience and work toward a trusted framework for agentic finance.

For banks, fintechs and payment providers, the message is not to slow innovation for its own sake. It is to keep systems understandable, accountable and resilient as their capabilities expand. As AI moves from offering suggestions to taking action, trust will depend on whether firms can show not just what their systems can do, but what they are authorized to do and how failures get handled.