How Trust Breaks Down in Practice

A. Where Trust Breaks Down: Operational Realities

The analysis begins with a distinction that recurred throughout the Executive Table: financial inclusion is not equivalent to digital confidence. Participants were clear that access alone is insufficient. A user may hold an account, transact digitally, and still remain uncertain whether the system will protect them when something goes wrong. Trust, in that sense, is not simply an attribute of products. It is an operating condition of the system itself.

Structured polling reinforced this point. Participants ranked transaction failure / service reliability and dispute resolution and recourse as the most significant trust breakdown points in the ecosystem, ahead of fraud, pricing opacity, KYC frictions, data privacy, and artificial intelligence. The ranking is analytically significant. It suggests that the most immediate drivers of confidence erosion are not the most technologically complex ones. They are the failures users encounter directly: payments that do not complete, charges that cannot be explained, losses that cannot be traced, and complaints that do not reach timely resolution.

Table 2  |  Trust Breakdown Points in Nigeria’s Digital Financial Ecosystem

Source: Bridgforte Executive Table participant polling, Lagos, February 2026. Scores reflect relative prioritisation across thirty participants and are indicative of ranking rather than precise magnitudes.

What matters for confidence is not simply whether systems fail, but how they behave when they do. Participants repeatedly returned to the view that failures are inevitable in complex financial systems; the critical determinant of trust is whether resolution is clear, timely, and fair. One participant cited Nigeria’s Central Securities Clearing System as evidence that reliability and high transaction volumes can coexist. Yet in large parts of retail digital finance, when failures occur, the experience of recourse remains uneven – and it is that experience, amplified by social media, that shapes confidence.

Three related dynamics emerged strongly from the discussion.

Asymmetric Perception Dynamics

The majority of transactions may complete without incident, yet negative experiences shape public perception disproportionately. Customer satisfaction, however, is typically unexpressed, while frustration is amplified through social media, messaging platforms and word of mouth. Confidence is therefore more sensitive to visible failure and poor recourse than to aggregate performance statistics.

“Satisfaction is structurally silent; frustration is structurally viral.”

Cash as the Enduring Competitor

Despite the growth of digital payments, cash remains the formal financial system’s most resilient competitor. This is not simply because cash is familiar, but because it offers certainty: immediacy, visibility, and control. In practice, cash functions as a benchmark against which digital trust is judged. Where fraud exposure, unexplained charges, or unresolved failures persist, users revert to cash not because digital channels are unavailable, but because they remain conditionally trusted.

The Informal Trust Paradox

Participants also pointed to the persistent strength of informal finance. Informal providers often command greater trust than formal institutions despite offering fewer legal protections and less sophisticated products. The explanation lies less in technical capability than in relational design: familiarity, responsiveness, empathy, and clear, socially embedded pathways for recourse. Formal systems have not yet translated these features into credible digital equivalents. This highlights Cultural Confidence as a structural dimension of trust, not a behavioural anomaly.

B. Identity Infrastructure as the Structural Root

Identity infrastructure emerged as the most consistently cited determinant of trust across the discussion. Participants identified fragmented KYC processes, inconsistent NIN implementation, and the inability to trace accounts to verified individuals as factors that simultaneously enable fraud, weaken consumer recourse, and diffuse institutional accountability. That breadth of impact, spanning fraud prevention, consumer protection, and institutional governance at the same time, makes identity the highest-leverage point for system-level intervention.

A key policy implication followed from this diagnosis: identity should be treated as a public good, not an institutional compliance cost. When identity functions as shared infrastructure rather than as a fragmented institutional process, it improves fraud prevention, reduces onboarding friction, strengthens consumer recourse, and clarifies accountability. Without reliable identity infrastructure, the financial system is constructing sophisticated products on an unstable foundation.

Nigeria possesses important components of a national identity architecture but has not yet achieved the interoperability and operational reliability required to deploy it as a trust anchor across institutions. The BVN and NIN systems represent significant foundational investments. The policy challenge is not their existence but their integration: the creation of a seamless verification environment that institutions can rely upon without duplicating effort or accepting inconsistent results. This is, fundamentally, a transition from asset creation to system integration.

A related observation concerned consumer perceptions of data privacy. Participants noted that many consumers express caution about sharing personal data with financial institutions while disclosing extensive information on social media platforms. This suggests that the trust challenge around data is not simply about disclosure, but about perceived value and control. Consumers will share data if they believe there is clear benefit and they retain meaningful agency over its use. The design of privacy practices – how institutions explain data use, what control mechanisms they offer, and how they respond to breaches – matters as much as the substance of those practices. Trust in data privacy is therefore shaped not only by policy, but by institutional behaviour.

Identity, in this sense, is not just an onboarding matter. It is a trust anchor.

C. Instant Payments: A Structural Design Tension

Real-time payments have been one of the major achievements of Nigeria’s digital financial transformation. But they also reveal one of its central design tensions: the very qualities that make instant payments attractive – speed, immediacy, irrevocability, and convenience – also increase the consequences of failure when safeguards do not evolve alongside the infrastructure.

The challenge is not whether to pursue fast payments. Their inclusion and efficiency benefits are well established. The challenge is how to embed risk management within payment architecture without undermining the speed that drives adoption. This is why the issue is best understood not as a product problem, but as a system-design problem.

The fraud trend illustrates this clearly. Losses rose from ₦12.7 billion in 2021 to ₦17.67 billion in 2023, then spiked to ₦52.26 billion in 2024, driven largely by a single ₦31.1 billion incident, before falling to ₦25.85 billion in 2025 following coordinated industry action. The scale of that outlier exposed how much damage a single point of failure can impose on a fast-moving system. Even setting aside the 2024 outlier, the pattern remains clear: fewer recorded incidents, but more damaging ones. The system’s fraud problem is less about volume than consequence. Social engineering, account compromise, and fraudulent merchant transactions – fraud vectors structurally enabled by the speed and irrevocability of instant payments – have displaced higher-volume, lower-value attacks.

This matters because instant payments compress the time available for intervention. Once funds move beyond recall, losses are realised quickly and often irreversibly. Participants discussed a number of possible design responses, including behavioural anomaly detection, transaction velocity limits calibrated to risk, and narrowly targeted delays where defined thresholds are triggered. The common principle was proportionality: risk controls should be embedded in the architecture itself, not layered on as blunt friction after the fact.

D. Artificial Intelligence: Trust Amplifier and Systemic Risk

Artificial intelligence featured in the Executive Table less as a present crisis than as an emerging governance frontier. Discussion was anchored by a plausible scenario: a fabricated AI-generated video purporting to show the chief executive of a major Nigerian bank warning of liquidity stress. The scenario mattered not because it was dramatic, but because it illustrated how confidence could be destabilised system-wide without any underlying institutional weakness.

The issue is not simply technological capability. It is the asymmetry between the falling cost of generating convincing synthetic content and the rising institutional burden of verification. Without rapid verification mechanisms – technical, institutional, and regulatory – misinformation alone could trigger systemic financial stress. Where systems for verification, contestability, and coordinated response are weak, AI can amplify trust shocks even before it becomes deeply embedded in financial decision-making.

Current Perceptions Among Senior Practitioners

Seventy-two per cent of participants selected “too early to assess” when asked about AI’s current trust impact – a distribution that confirms the watchful rather than crisis framing. Most did not describe AI as already eroding trust directly. Instead, they located the principal risks in governance failure: misuse of data, lack of explainability, biased decision-making, and inadequate mechanisms for challenge and redress. That indicates a limited but important governance window. Systems have not yet fully absorbed the consequences of AI deployment, which makes this a moment for institutional design rather than reactive damage control.

Table 3  |  Greatest Trust Risks from AI in Financial Services

Source: Bridgforte Executive Table participant polling, Lagos, February 2026. Figures reflect the share of respondents selecting each category.

Participants located the primary AI risks in governance – how AI is overseen, explained, and contested – rather than in the technology itself. Nine in ten participants located the greatest risk in data misuse, explainability failures, or embedded bias. The appropriate response is therefore not to slow adoption indiscriminately, but to strengthen the governance architecture around it.

The Dual-Use Challenge

The same algorithmic capability that powers fraud detection can also enable deepfake video generation. The same credit-scoring models that extend access to underbanked consumers can also embed and scale historical bias present in training data. Governance frameworks must address both sides of this duality. They cannot be designed only to promote beneficial applications; they must also constrain harmful ones. The discussion suggests that governance, rather than capability, is the defining challenge of AI deployment in financial systems.

As one participant observed, yesterday’s inequities can become tomorrow’s infrastructure unless institutions actively intervene to correct for historical bias in training data. This points to a governance requirement that extends beyond explainability. It requires active monitoring, regular auditing, and a willingness to modify or retire models whose real-world effects diverge from design intent. The challenge is not whether AI is beneficial or harmful. It is whether its deployment is governed well enough to preserve confidence.

The Board Governance Deficit

Three diagnostic questions emerged as a practical test of institutional AI readiness. Do boards exercise explicit oversight of AI model performance, including bias testing, performance drift monitoring, and explainability standards? How are institutions monitoring AI’s differential impact across demographic segments? How are automated decisions explained to affected individuals, and through what mechanisms can those decisions be challenged?

The candid assessment in the room was that most boards are not yet equipped to answer these questions with confidence. The gap between AI deployment and AI governance is not a technical gap; it is a structural one, and it is widening.

E. The Collaboration Paradox: Structural Barriers to Collective Defence

Perhaps the clearest structural insight to emerge from the Executive Table was that financial institutions often collaborate less effectively than the fraud networks targeting them. The cause is structural rather than moral. Criminal networks face no competitive barriers to information sharing. Financial institutions face many: competitive instinct, reputational risk, regulatory uncertainty around data exchange, and legal ambiguity about the permissibility of information-sharing arrangements. The result is a systemic weakness in Ecosystem Coordination rather than a series of isolated institutional failures.

The consequence is a structural asymmetry. Individual institutions invest heavily in their own defences while collective vulnerabilities continue to grow. Fraud intelligence that could allow one institution to protect many is instead siloed. Threat patterns that have appeared at one bank re-emerge at others weeks or months later because early warning signals are not shared quickly enough, or at all.

Table 4  |  Barriers to Collaboration in Nigeria’s Financial Ecosystem

Source: Bridgforte Executive Table participant polling, Lagos, February 2026. Scores reflect relative prioritisation across thirty participants and are indicative of ranking rather than precise magnitudes.

The ranking is revealing. Institutional mistrust and competitive incentives dominate the barrier landscape, while regulatory uncertainty, legal constraints, cost, and technology rank materially lower. The primary barriers to collaboration are therefore relational and structural, not legal or technical. Regulatory reform alone will not resolve them. They require institutional design: trust-building mechanisms, neutral intermediaries, and in some cases mandatory participation frameworks. The binding constraint is not capability, but coordination architecture.

The Bank-Fintech Fault Line

A specific collaboration gap was identified as particularly consequential: the persistent disconnect between banks and fintechs. Fintech-to-fintech and bank-to-bank coordination function reasonably well within their own categories. Bank-to-fintech collaboration remains much weaker, even though these institutions increasingly share operating environments, customer populations, and infrastructure. This fragmentation weakens both innovation efficiency and system resilience. Trust failures increasingly emerge at the intersections between institution types, not only within them.

Proven Precedents for Shared Infrastructure

Nigeria’s most effective domestic trust infrastructure – including BVN, NIBSS, and the GSI – was built through cooperative investment rather than competitive differentiation. At the continental level, Nigeria’s active participation in PAPSS reflects the same principle applied to cross-border payments. Each emerged from recognition that some infrastructure is more valuable as a shared utility than as a proprietary asset. The case for applying the same logic to fraud intelligence, cybersecurity capacity, and AI training data is therefore not merely theoretical; Nigeria’s own track record suggests it is viable.

The Nigeria Electronic Fraud Forum (NeFF) has demonstrated since 2011 that cross-institutional fraud collaboration, when properly convened and backed by regulation, can produce measurable results. Open banking infrastructure, whose framework and operational guidelines are finalised and awaiting CBN go-live confirmation, will establish standardised data-sharing rails between banks and licensed third parties. But NeFF is advisory and voluntary, and open banking addresses data interoperability rather than the broader governance of shared threats. The financial system still lacks a standing cross-sector body with a mandate spanning all institution types and addressing trust as a systemic challenge beyond any single domain.

Concrete proposals emerging from the discussion included a National Cybersecurity Centre that would aggregate institutional budgets into shared continuous monitoring and threat response capability; an anonymised fraud data-sharing platform, potentially using distributed ledger technology, overseen by a neutral body with regulatory authority; and shared AI training datasets that would improve model accuracy while reducing cost burdens on individual institutions.

These precedents show that coordination failures are not inevitable. They are, at least in part, design choices. Collaboration must therefore be designed, not assumed.

F. Reimagining the Regulatory Relationship

A recurring theme in the Executive Table was the changing nature of the relationship between institutions and regulators. Participants did not primarily describe a crisis of legitimacy. Rather, they described a growing mismatch between supervisory models designed for discrete institutions and systems in which risk emerges through interaction, interdependence, and technological change.

Institutions often reported feeling heavily engaged on process, but less engaged on the underlying system questions that matter most: how accountability should be assigned across multi-party failures, how AI deployment should be governed, how regulatory dialogue should function in fast-moving areas of innovation, and how systemic risk should be addressed where no single actor fully contains it.

What emerged was a call for a philosophical shift in supervisory approach: from mandate-based regulation, which prescribes how institutions must operate, to outcome-based supervision, which defines what results the system must achieve while allowing greater flexibility in how institutions achieve them. This implies a move from controlling institutional behaviour to shaping system outcomes. Participants argued that innovation should help inform where regulation needs to go, rather than waiting for regulation to define the boundaries within which innovation is permitted.

At the same time, there was clear recognition of the constraints regulators face. Innovation is moving faster than regulatory understanding in many technological domains. Technical expertise is unevenly distributed within supervisory institutions. The volume of entities and products requiring oversight continues to expand faster than supervisory capacity. Participants were divided on the implications: some saw this capacity gap as grounds for greater industry self-regulation in defined areas; others argued that the regulator’s mandate for systemic stability gives both the authority and the obligation to build capacity rapidly.

This capacity gap strengthens the case for structured dialogue: regulators need access to practitioner knowledge, and practitioners need regulatory engagement that is substantive rather than purely compliance-oriented. A practical proposal that commanded broad support was the creation of standing mechanisms between operators and regulators, not as crisis-driven consultations but as regular channels for substantive engagement. Such dialogue requires a neutral convener trusted by all parties. Sustained coordination between regulators and operators is itself part of trust architecture.

Clear liability frameworks for cross-institutional failures are a prerequisite for effective consumer protection. Where fraud originates in one institution, passes through another, and settles with a consumer served by a third, the current architecture often produces fragmented accountability rather than a unified response. When responsibility cannot be explained across a multi-party transaction, redress becomes harder to secure and confidence weakens accordingly.

The Overall Trust Assessment

The findings across each of these dimensions provide the context within which the room’s overall assessment must be read. After working through the specific failure dimensions discussed in this section, participants were asked to rate trust resilience in Nigeria’s financial ecosystem on a scale of 1 to 10. The average score was 5.4. Scores clustered between 3 and 8, with the largest single group selecting 5.

A middling score might appear inconclusive. It is not. These are the chief executives, board directors, and senior institutional leaders closest to the system’s performance and most invested in its credibility. Their collective assessment, after considering the failure dimensions surfaced in the discussion, was neither confidence nor alarm, but caution. In a system that depends on confidence, such caution speaks for itself.

The score therefore carries more than descriptive value. It reflects a measured and guarded assessment from the senior cohort responsible for Nigeria’s financial infrastructure: a recognition that the system is capable, but that its confidence architecture remains uneven.

Synthesis

Taken together, these themes point to a common conclusion. The most important trust failures in Nigeria’s digital financial ecosystem are not isolated incidents or purely institutional weaknesses. They are interdependent failures arising at the intersection of infrastructure, accountability, governance, and coordination. This is why trust cannot be treated as a by-product of institutional performance alone. It is shaped by the system-level conditions through which institutions interact, users experience the system, and failures are resolved.

The next section turns from analysis to implication: what these findings mean for the design, governance, and regulation of digital financial systems.