AI is becoming part of everyday life across asset management, and the Financial Conduct Authority (FCA) has a consistent message for firms using it: pair AI innovation with strong governance, transparency, resilience, and accountability to ensure safe and responsible adoption.

The FCA recognises that AI is already embedded across financial services, with most firms exploring or deploying AI-enabled tools. It sees real opportunities for firms, particularly through agentic AI systems that go beyond analysis and automation to support decision-making, coordination, and execution of financial activities. Combined with developments such as tokenisation and programmable financial infrastructure, AI has the potential to reshape investment operations, client servicing, and the way markets function.

At the same time, the FCA wants to make sure innovation does not undermine trust in financial markets. Its central expectation is that responsibility for regulated activities stays with firms and their senior managers, however sophisticated AI systems become. Firms cannot outsource accountability to an algorithm. Investment decisions, client outcomes, and regulatory obligations still need appropriate human oversight and challenge.

What asset managers should focus on

1. Keeping accountability with humans

The FCA accepts that AI will increasingly support portfolio management, research, client servicing, and operational processes. What it will not accept is a “the black box made the decision” defence. Senior managers and firms remain accountable for regulated activities and client outcomes, which means they need:

• Clear ownership of AI-enabled processes

• Defined escalation and override procedures

• Evidence that people can understand and challenge AI outputs

• Appropriate oversight from boards and investment committees

2. Treating AI as a resilience issue, not just an innovation project

The FCA keeps coming back to operational resilience and third-party dependencies. In practice, that means firms should:

• Map their reliance on cloud providers, model providers, and data vendors

• Assess concentration risk

• Maintain contingency plans for outages or model failures

• Build AI dependencies into operational resilience testing

The regulator expects firms to know where critical AI services sit in their infrastructure, and how they would manage a failure.

3. Strengthening governance and risk management

AI governance is now a board-level issue. Good practice increasingly includes:

• Board oversight of AI strategy and risk appetite

• Formal AI policies and approval processes

• Model validation and performance monitoring

• Bias, fairness, and explainability assessments where relevant

• Regular reporting to senior management

4. Using AI to improve market integrity

Interestingly, the FCA is using AI itself to monitor wholesale markets and detect market abuse faster. Firms should expect growing regulatory interest in how they use AI to strengthen surveillance, fraud detection, compliance monitoring, and risk management, not just to cut costs.

5. Readying for closer scrutiny of system-wide risks

As more firms adopt similar AI tools and data sources, the FCA is paying closer attention to system-wide risks and competition issues. Some key questions to consider include:

• Could widespread use of the same models create correlated behaviour across the market?

• How well governed are data-sharing arrangements?

• Could AI adoption affect market resilience during a period of stress?

Recent developments to watch

The FCA's position has moved from broad principles towards more concrete supervisory expectations over the past few months, and asset managers should keep an eye on the following:

• In March 2026, the FCA published its first Wholesale Buy-Side Regulatory Priorities Report, replacing the old portfolio letters. It expects firms using AI, distributed ledger technology, or tokenised structures to have clear governance, oversight, and risk management in place, and it specifically flagged buy-side firms' reliance on third-party providers as a source of concentration risk.

• In May 2026, the FCA reopened its AI Input Zone, seeking industry views on what good AI practice looks like, and the FCA, the Bank of England, and HM Treasury issued a joint statement urging firms to strengthen cyber resilience against risks from frontier AI models.

• In June 2026, the FCA reaffirmed that it does not intend to introduce AI-specific rules for financial services, and will instead supervise AI through existing frameworks such as the Consumer Duty and the Senior Managers and Certification Regime (source FCA).

None of this changes the underlying message from the FCA, but it does show it moving from principle to practice more quickly than some firms may have expected.

The practical takeaway

The FCA's emerging position can be summed up as “innovate, but stay accountable”. Firms do not need to wait for a dedicated AI rulebook before acting. The regulator already expects firms to apply existing governance, operational resilience, consumer protection, and market integrity principles to AI-enabled activities.

The firms that will be viewed most favourably are those that can demonstrate:

• A clear AI strategy

• Strong governance over AI processes

• Appropriate human oversight

• Robust resilience planning

• Thoughtful management of third-party dependencies

For asset managers, the question is no longer whether to adopt AI, but how to do so in a way that stays consistent with fiduciary responsibilities and regulatory obligations. Firms that combine innovation with strong governance will be best placed to benefit from the opportunities ahead.