AI governance is fast becoming one of APAC’s most critical leadership challenges and trust is central to getting it right because without it, users will avoid integrating the tool into their workflow operations. Randy Goh, Area Vice President, ASEAN, Dataiku, explores why AI governance is emerging as one of the region’s key challenges – and how enterprises can close the gap between ambition and execution.
AI is being deployed globally at unprecedented speed – but it’s becoming increasingly difficult to keep pace with governance, trust and operational readiness. This growing imbalance is fast turning into a systemic enterprise and economic risk, particularly in Asia Pacific.
According to the World Economic Forum’s Advancing Responsible AI Innovation report, despite rising awareness, responsible AI practices remain largely immature. A 2025 survey of 1,500 organisations found that 81% are still in the first two early stages of responsible AI maturity. In Asia Pacific specifically, the gap is even more pronounced.
One of the key themes throughout was the importance of strong governance and building trust, as well as the need to invest in the underlying infrastructure that allows AI to scale effectively, beyond just data centres.
Randy Goh, Area Vice President, ASEAN, Dataiku, offers his expert advice.
Why is AI adoption outpacing governance and what systemic risks does this create for enterprises in APAC?
Firstly, we have to evaluate Artificial Intelligence (AI) not just from a strategy perspective, but also from a governance, data readiness and human talent perspective.
The reality is that AI adoption in Asia Pacific (APAC) is accelerating faster than internal governance mechanisms can keep pace. This leaves enterprises exposed to systemic risks, including shadow AI, data misuse, biased decision‑making and opaque incident accountability across borders.
The full extent of the matter was laid bare in the recent World Economic Forum Advancing Responsible AI Innovation report, which found that only 1% of organisations in the region have fully operationalised responsible AI, with most relying on partial, defined but limited, or ad hoc measures.
Even in Singapore, since the launch of the National AI strategy 2.0 (NAIS 2.0) and Smart Nation 2.0 in 2023 and 2024, robust enterprise-level controls, policies and monitoring remain elusive for many organisations.
Further exacerbating the issue for enterprises in APAC is the absence of binding regional legislation, making it more challenging to realise interoperability and cross-border technology integration within an enterprise.
Why is trust now considered a core component of AI success, not just a compliance requirement?
Trust has moved from being a narrow regulatory concern to a fundamental driver of whether AI delivers real value at scale. The success of AI adoption depends less on technical capability alone and more on whether people believe the systems are safe, fair and reliable. Without that belief, productivity gains and innovation simply do not materialise. Public sector efforts such as Singapore’s national AI initiatives emphasise that workers and citizens must feel confident and discerning when using AI. Only when people trust these tools will they integrate them into daily decisions and workflows, thereby generating broad economic benefits.
When trust is lacking, fear quickly replaces curiosity. Concerns about constant monitoring, biased outcomes, or misuse of personal data lead employees and citizens to disengage from AI systems or actively circumvent them. In fact, according to Dataiku’s Global AI Confessions Report: Data Leaders Edition, ‘75% of data leaders say trust in their AI agent deployments is a concern’. This behaviour erodes the expected return on AI investments and slows organisational transformation.
Trust also plays a critical role in enabling access to high-quality data, which is the lifeblood of effective AI. Policy frameworks such as Singapore’s NAIS 2.0 stress the importance of trusted cross-border data flows and the use of privacy-enhancing technologies. These mechanisms allow organisations to collaborate safely and develop advanced AI applications in areas like healthcare, financial services and supply chains. In contrast, when stakeholders doubt how data will be governed, organisations become protective of their most valuable datasets. This reluctance restricts model accuracy, limits innovation and reduces the practical impact of AI solutions.
Beyond operational considerations, trust is now inseparable from strategic risk management. Research by bodies such as The Organisation for Economic Co-operation and Development (OECD) highlights that untrustworthy AI can negatively affect worker rights, safety and equality. High-profile failures can rapidly erode public confidence, not only in specific systems but also in the institutions that deploy them. For organisations operating across the Asia-Pacific region, where regulatory expectations differ by jurisdiction, relying solely on minimum compliance is increasingly risky. AI incidents that cross borders can trigger reputational harm, regulatory intervention and loss of public legitimacy. As a result, cultivating trust has become essential to sustaining long-term AI success, not merely avoiding legal penalties.
How can enterprises in APAC balance rapid AI innovation with the need for transparency, explainability and accountability?
Enterprises across the Asia-Pacific region face the challenge of innovating quickly with AI while also meeting growing expectations around openness, interpretability and responsibility. Achieving this balance requires moving beyond reactive compliance and adopting governance approaches that are built directly into AI development processes. Governance-by-design models allow organisations to move fast while still managing risk, even in a region characterised by diverse and evolving regulatory regimes. Agile, ongoing risk assessments ensure that innovation can continue without ignoring potential social, ethical, or operational consequences.
Frameworks such as Singapore’s FEAT principles, which focuses on fairness, ethics, accountability and transparency, demonstrate how voluntary self-regulation can support innovation rather than restrict it. By providing clear ethical direction without rigid rules, these principles enable companies to iterate rapidly while maintaining essential safeguards. This is particularly important in high-impact sectors like banking, financial services and insurance, where opaque or biased AI systems can quickly damage public trust and trigger regulatory attention. Embedding such principles early helps organisations prevent reputational harm before it occurs.
However, principles alone are not enough. To unlock real business value, responsible AI must translate into concrete operational practices. As industry research has highlighted, this means establishing strong governance structures, conducting regular risk and impact assessments and subjecting systems to thorough testing before and after deployment. Continuous monitoring is equally critical, ensuring models remain reliable and aligned with organisational values as conditions change over time. These practices make accountability tangible rather than symbolic.
Equally important is recognising the broader human and resilience dimensions of AI adoption. Enterprises must consider how AI affects employees, long-term workforce skills, environmental sustainability, data privacy and cybersecurity. Addressing these factors strengthens organisational resilience and signals a serious commitment to responsible innovation. Encouragingly, many APAC leaders are already investing heavily in Generative AI (GenAI) to develop new products and services.
Responsible AI serves as the bridge between experimentation and sustainable impact, allowing organisations to scale solutions securely, build lasting trust with stakeholders and convert ambitious AI strategies into durable business outcomes.
How can companies avoid the common ‘AI hype’ traps?
As enthusiasm for GenAI accelerates, many organisations fall into predictable hype-driven pitfalls that dilute value and increase risk. Rapid experimentation without adequate oversight can quickly create governance blind spots, leading to ethical concerns, hidden technical complexity, unapproved ‘shadow AI’ and systems that are difficult to explain or control. In fact, our Global Data Report showed that 94% of CEOs suspect employees are already using GenAI tools without notice or permission. To avoid these traps, companies need discipline, clarity and strong foundations alongside innovation.
The first step is to ground AI initiatives firmly in business priorities. Rather than pursuing the latest tools or trends, organisations should start with well-defined problems that AI is uniquely positioned to address. Successful programmes link AI use cases to tangible outcomes such as improved customer loyalty, higher productivity, or reduced operational and compliance risk. Defining clear outcome charters helps teams measure success in business terms, ensuring that technical achievements translate into real impact.
A second common failure point is fragmentation. When teams experiment independently without coordination, organisations accumulate overlapping models, inconsistent standards and unmanaged risk. Adopting a unified platform and enterprise-wide governance approach reduces this complexity. Centralised oversight allows companies to manage models, agents and workflows consistently while still enabling teams to innovate quickly. Effective governance frameworks should be flexible enough to support speed, while robust enough to meet regulatory and risk requirements across multiple markets.
Honest self-assessment is equally important. AI hype often disguises gaps in data quality, infrastructure, or skills. Many organisations assume they are more mature than they truly are, leading to stalled projects and unmet expectations. Conducting realistic maturity assessments across technology, governance and organisational culture helps identify constraints early and align AI adoption with long-term strategic objectives rather than short-lived experimentation. Strong data foundations are another critical safeguard against hype. Even the most advanced AI models cannot compensate for unreliable or poorly governed data. Building on secure, well-managed and locally compliant data sources is essential, particularly in regions with high sensitivity around data sovereignty and privacy. Treating data governance as a core capability underpins both trust and performance.
Finally, companies must move beyond isolated pilots. AI creates lasting value only when it is embedded into everyday workflows and decision-making. By integrating AI into routine operations and fostering a human-centred culture, organisations can ensure that AI enhances employee capabilities rather than remaining an experimental side project.
