What members should do next

Start with four actions: assign owners for AI controls, review fraud data sharing, identify gaps in guidance and raise questions about AI-led payment disputes.

Members can strengthen AI governance and intelligence sharing now.

Start with decisions within the firm’s control: identify AI systems used in fraud and AML, assign accountable owners and review monitoring and dispute handling. Map the fraud signals the firm shares and receives, then assess gaps in coverage, quality and response times.

Members can also identify practical questions that would benefit from clarification by government or regulators. Worked examples of governance problems and disputed payments can make those questions more precise. The recommendations below distinguish action within firms from issues requiring wider coordination.

Members should begin with decisions within their control: ownership of AI systems, evidence of control performance and the usefulness of shared fraud intelligence.

The policy questions require the same precision. Where practical guidance is needed, identify the use case and the decision that remains unclear. For disputed AI-initiated payments, distinguish questions about authority, execution and fraud.

The Payments Manifesto 2027 sets out the Association’s policy position. Read the survey findings alongside that position, rather than presenting them as proof that mandatory sharing is the only effective approach.

Simon Miller, director of policy, strategy and communications, Cifas

Interview takeaway: Disrupt scam content and criminal activity earlier, and assess detection quality as well as volume.

Four first steps, with owners and dates

These proposed actions give members a starting point for internal review and for identifying questions that require wider clarification.

Recommended action
Owner
First deliverable
Proposed deadline
Review AI used in fraud and AML controls
Accountable executive, supported by fraud, AML, technology and risk teams
Inventory of live and planned AI systems, their purposes, owners, suppliers, approvals, monitoring and escalation arrangements; a prioritised plan to address gaps
Within 90 days of publication. Assess new deployments before launch.
Review fraud intelligence sharing and investment
Accountable executive, supported by fraud, data, technology, legal and privacy teams
Map of signals, channels, data quality and participation gaps; costed options for improvement, including EDEx where relevant
At the next budget review, and no later than 31 March 2027
Identify practical AI guidance needs
Fraud and AML leads, supported by risk and compliance teams
Worked examples showing where explainability, monitoring or oversight expectations need clarification; consider a response to CP26/38 where relevant
Review before the consultation closes on 24 December 2026
Review AI-initiated payment dispute scenarios
Payments, fraud, legal and compliance teams at firms developing or using AI-initiated payments
Document authority, audit trails and dispute handling; identify specific questions requiring clarification
Before deployment; review existing arrangements at the next governance meeting

Measuring progress

Members should track whether material AI systems have documented owners and controls, whether identified gaps have remediation dates, and whether sharing improvements produces useful signals and faster responses. Members should also track relevant government and regulatory responses and review what they mean for their own controls and investment plans.

Intelligence sharing: from support to investment

Previous page