Agentic AI systems are already executing financial workflows autonomously in UK organisations — processing payments, reconciling accounts, and routing transactions with minimal human oversight. The FCA, in its March 2026 Payments Regulatory Priorities report, placed agentic AI squarely in its sights: existing payment rules were not written for autonomous systems, and accountability questions around liability, consent, and control remain unresolved. For UK CFOs, this is no longer a technology question. It is a governance and legal accountability question — one that lands directly on the finance director’s desk.

What Is Agentic Finance and Why Does It Matter for UK CFOs Right Now?

Agentic AI refers to autonomous systems capable of independently executing multi-step financial workflows without human intervention at each stage. Unlike traditional automation, which follows pre-defined rules, agentic systems use large language models to reason, adapt, and act — initiating payments, forecasting cash positions, closing the books, and flagging anomalies in real time.

According to research from ChatFin, the global agentic AI spending market reached $12.4 billion in 2026, with 76% of CFOs now allocating dedicated budgets for autonomous finance agent deployment. The evaluation phase is largely over. What was a pilot in 2025 is live infrastructure in 2026.

In the UK, the convergence of rapid adoption and regulatory lag creates a specific risk for finance leaders. Current payment services legislation — rooted in the Payment Services Regulations 2017 — was designed for human-authorised transactions. Where an AI agent initiates or routes a payment autonomously, the legal framework for liability, consent, and audit trails is ambiguous. The FCA has flagged this explicitly. Firms cannot wait for formal rule clarification before establishing internal accountability structures.

Executive Action

• Map every autonomous finance agent in your organisation and document its decision boundaries and escalation triggers.
• Review your Payment Services Regulations 2017 obligations specifically in the context of AI-initiated transactions.
• Brief your General Counsel on the FCA’s March 2026 agentic AI signals now — do not wait for formal rule changes to begin your response.

What Does the FCA Expect from CFOs Deploying Autonomous Finance Agents?

The FCA’s March 2026 Payments Regulatory Priorities report was unambiguous: firms deploying agentic AI in payments must demonstrate effective oversight and control. Specifically, the regulator noted that where an individual is responsible for an agentic AI system, they must maintain “an appropriate level of control” — a requirement that, in the FCA’s own framing, “seemingly counters the autonomous nature of these models.”

This tension is the central governance challenge of agentic finance. An AI agent that truly acts autonomously, by definition, removes human decision-making from the loop. If the regulator requires human control, organisations must define what meaningful oversight looks like — without nullifying the efficiency gains that justified the investment in the first place.

Under the Senior Managers and Certification Regime (SMCR), responsibility for material financial systems and controls sits with named senior managers. AI-driven finance infrastructure almost certainly falls within scope. CFOs should assume that the FCA will treat agentic AI failures — erroneous payments, regulatory breaches, or financial crime facilitation — as SMCR accountability events. The UK faces an annual £195 billion money laundering exposure according to Treasury analysis; autonomous agents operating at machine speed could dramatically amplify that risk if governance structures are not in place.

Executive Action

• Assign a named SMCR senior manager as accountable for every agentic finance system in production — this is not optional.
• Establish real-time audit logs for all AI-initiated transactions; the FCA will expect to inspect these under existing supervisory powers.
• Review contractual arrangements with AI vendors for liability allocation and indemnification clauses specific to autonomous transaction errors.

How Should UK CFOs Build a Governance Framework for Autonomous Finance?

The IMF’s 2026 working paper on agentic AI and payments identifies three governance pillars that apply directly to UK finance leaders: transparency (can the system explain its decisions?), auditability (can regulators trace every action?), and reversibility (can erroneous actions be unwound quickly?). Building these three pillars into your agentic finance deployment is not a best practice — it is the baseline that FCA and SMCR scrutiny will demand.

Senior finance leaders should treat agentic AI governance as an extension of existing internal controls frameworks, not a standalone technology project. INFORMD has been tracking the FCA’s evolving position on AI in financial services throughout 2026, and the clearest signal from regulators is consistent: the sophistication of the tool does not diminish the accountability of the executive. Boards should receive regular reporting on agentic AI performance, exceptions, and near-misses in the same cadence as other material risk indicators.

Practically, this means updating three documents most CFOs already own: the risk register (add agentic AI as a named risk category), the internal controls framework (define control objectives and exception thresholds for autonomous systems), and the board reporting pack (add an AI operations summary covering volume, overrides, and regulatory flags).

Executive Action

• Incorporate agentic AI into your internal controls framework with explicit reference to Provision 29 obligations effective for financial years beginning 1 January 2026.
• Add AI operational performance — volume, exceptions, overrides — to board reporting at least quarterly.
• Commission an independent audit of any agentic finance system before scaling beyond pilot or extending to customer-facing payment flows.

What Are the Compliance Risks UK Finance Executives Cannot Ignore?

Beyond the FCA, UK CFOs deploying agentic AI face a layered compliance environment. Under UK GDPR, automated decision-making that produces legal or significant effects requires a lawful basis and, in many cases, the right to human review. An AI agent autonomously declining a payment, freezing a supplier account, or flagging a transaction for investigation may trigger Article 22 obligations — an issue few finance teams have stress-tested against live deployments.

The Financial Reporting Council’s reforms, effective for financial years beginning 1 January 2026, require boards to formally declare on the effectiveness of material controls under Provision 29. If agentic AI is embedded in material financial processes — and for most mid-to-large organisations that deploy it, it will be — that declaration must account for it explicitly. Boards that sign off without examining their AI infrastructure face reputational and regulatory exposure that will be difficult to contain after the fact.

The governance gap between technology deployment and executive accountability is closing fast. Those who act now — mapping systems, assigning ownership, and building audit infrastructure — will be in a materially stronger position when the FCA publishes updated payment services rules, which are expected later in 2026.

Executive Action

• Conduct a UK GDPR Article 22 assessment for every agentic AI system that makes or influences decisions affecting counterparties, suppliers, or customers.
• Ensure agentic AI systems and their material control implications are reflected in your Provision 29 board declaration for 2026.
• Engage with INFORMD’s executive briefings via our resources library to stay current as FCA rule changes emerge — the pace of regulatory movement in this area is accelerating.

INFORMD provides intelligence briefings for senior business leaders across technology, finance, strategy, and compliance. Based in Milton Keynes, UK, we help executives stay informed and act with confidence. Explore our full library of executive briefings or speak to our team.

Frequently Asked Questions

What is agentic AI in finance?

Agentic AI in finance refers to autonomous systems that independently execute financial tasks — processing payments, closing accounts, reconciling ledgers, and forecasting cash flow — without requiring human approval at each step. Unlike rules-based automation, these systems can reason and adapt to changing conditions, making real-time decisions across complex workflows.

Is the FCA already regulating agentic AI in payments?

The FCA addressed agentic AI directly in its March 2026 Payments Regulatory Priorities report. While formal rule changes have not yet been published, the regulator is clear that firms must maintain effective oversight and control of autonomous systems today — under existing payment services legislation and SMCR frameworks. Formal rule updates are expected later in 2026.

Who is legally accountable if an AI agent initiates an erroneous transaction?

Under SMCR, a named senior manager is accountable for material financial systems and controls. If an agentic AI system initiates a transaction that breaches regulations or causes financial harm, the FCA will expect to identify a responsible individual — most likely the CFO or a designated executive with formal oversight of finance infrastructure. Vendor contracts do not transfer this regulatory accountability.

How should UK CFOs prepare their boards for FCA scrutiny of agentic finance systems?

CFOs should document every autonomous finance system in operation, assign a named SMCR-accountable executive, establish audit trails for AI-initiated decisions, and incorporate agentic AI risks into the internal controls framework and Provision 29 board declarations. Regular board reporting on AI performance and exceptions should begin immediately — before the FCA’s updated rules arrive.

Stay ahead. Subscribe to INFORMD’s executive briefing at informd.co.uk/services.