Between innovation and precaution: what Indonesia’s emerging AI governance framework signals for Asia

Across Asia, from Japan to India, countries are increasingly moving from conversations about artificial intelligence adoption towards questions of governance. Indonesia’s recent announcements may therefore represent more than another national AI initiative.

Through calls for stronger ethical oversight in healthcare and proposals for an umbrella framework supported by sector-specific regulations, the country is beginning to grapple with a challenge confronting policymakers across the region: not whether AI should be embraced, but how institutions can ensure that innovation remains accountable to the societies it seeks to serve.

The answers Indonesia arrives at may ultimately offer important lessons for Southeast Asia as the region determines the conditions under which AI should operate.

History’s lessons: innovation has always tested institutions

Technological progress has rarely unfolded alongside equally rapid institutional adaptation.

The Industrial Revolution transformed economies but exposed workers to unsafe conditions, eventually giving rise to labour protections. The thalidomide tragedy reshaped pharmaceutical regulation, reinforcing the need for independent scientific oversight. Even the early internet evolved faster than the rules governing privacy, competition, and digital rights.

The lesson is not that innovation should be feared. Rather, institutions rarely evolve at the pace of technology. The challenge for policymakers is to build governance before harm compels reform.

Healthcare AI presents precisely such a moment.

The Indonesian approach: a middle path?

Against this backdrop, Indonesia’s approach deserves careful attention.

Deputy Minister of Health Dr. Dante Saksono Harbuwono asked a deceptively simple question: “When a computer program recommends a diagnosis to a doctor, who is responsible if the recommendation is wrong?”

The significance lies in the shift from performance to accountability. In healthcare, algorithmic errors are not merely technical failures; they can directly affect patient outcomes.

Dr. Dante’s observation that: “Innovation without governance is risk. Governance without innovation is stagnation.” may be among the most important policy signals emerging from the region.

It rejects two unhelpful extremes: the assumption that regulation inevitably suppresses innovation, and the belief that innovation should proceed unchecked until harms emerge.

Indonesia’s proposals suggest an attempt to navigate a middle path. Plans to regulate AI-enabled medical devices, strengthen patient consent frameworks, and expand ethics oversight indicate an understanding that healthcare requires differentiated governance.

This perspective is reinforced by remarks from Minister of Communication and Digital Affairs Meutya Hafid, who noted that Indonesia’s legal framework will prioritise ten sectors, including health, education, finance, energy, and transportation. The statement confirms a sector-based approach to AI governance rather than a single, all-encompassing law.

This distinction matters because healthcare is not simply another sector. When algorithmic errors can directly affect patient outcomes, clinical accountability, patient safety, and public trust require governance frameworks that extend beyond technological performance alone.

Science and policy: a delicate relationship

Effective governance depends on the relationship between scientific evidence and policymaking.

Researchers may ask whether an algorithm is accurate and generalisable. Governments must additionally consider affordability, scalability, and workforce readiness. Businesses assess commercial viability. Citizens ask a different question altogether:

Can this system be trusted?

These perspectives need not conflict, but tensions emerge when one dominates the others.

Scientific evidence must remain sufficiently independent to preserve credibility, while policymakers must interpret that evidence within broader social and economic realities.

Healthcare illustrates this balance particularly well. An algorithm performing well under controlled conditions may prove unsuitable within resource-constrained settings lacking infrastructure or trained personnel. Conversely, delaying innovation indefinitely may deny populations access to beneficial technologies.

The challenge is not insulating science from policy, but ensuring evidence informs decision-making without becoming captive to political expediency or commercial interests.

What this means for ASEAN

Indonesia’s announcements extend beyond national borders.

ASEAN is characterised by significant diversity in health system maturity, digital infrastructure, and regulatory capacity. Within this landscape, Indonesia’s emerging framework may offer important lessons.

First, the proposed “umbrella plus derivative regulation” model demonstrates how broad principles can coexist with sector-specific implementation.

Second, it could stimulate regional discussions around interoperability, cross-border data governance, and AI-enabled medical devices.

Finally, Indonesia may become a test case for whether context-sensitive governance can balance developmental ambitions with public-interest safeguards.

Too often, AI governance debates are framed as choices between importing highly prescriptive models or embracing market-led experimentation. Southeast Asia’s realities suggest a third possibility: governance frameworks rooted in local contexts while remaining responsive to global norms.

The real challenge, however, lies in implementation.

Public interest beyond ethics

Indonesia’s announcements also deserve recognition for elevating public trust and data governance within the broader AI conversation.

References to patient consent, retinal data collection, the role of SATUSEHAT AI, and concerns surrounding participation in data-sharing initiatives suggest that policymakers are grappling with a more fundamental question: under what conditions are citizens willing-and able-to entrust institutions with their data?

This is encouraging.

Public trust cannot be assumed simply because a technology demonstrates technical efficacy. Healthcare systems rely on legitimacy. People disclose sensitive information because they believe institutions will use it responsibly and remain accountable when harms occur.

This requires moving beyond ethics as a checklist.

Equally significant was Minister Meutya Hafid’s acknowledgement that authorities were “surprised by the large number of people who flocked to provide data to receive incentives.” The observation highlights an often-overlooked dimension of digital health governance: meaningful consent cannot be reduced to the act of agreeing.

Economic vulnerability, limited data literacy, and unequal access to information can undermine meaningful consent. The challenge for policymakers is not merely to secure agreement, but to ensure that participation is informed and genuinely voluntary.

Digital systems may also inadvertently exclude already vulnerable populations, making mechanisms for redress and accountability essential. Transparency, meanwhile, extends beyond technical documentation; people must be able to understand how decisions affecting their care are being shaped.

Public trust should not be treated as an outcome of governance. It should be one of its design principles.

The sustainability dimension

Another dimension warrants greater attention: sustainability.

Current AI discussions frequently focus on fairness, bias, and accountability. Yet the infrastructures supporting AI depend on energy systems, water resources, and increasingly complex supply chains.

Recent United Nations estimates suggest that data centres supporting AI could consume water equivalent to the needs of 1.3 billion people by 2030. These concerns are not disconnected from healthcare.

AI-enabled diagnostics and digital health platforms can improve efficiency and access, but these gains remain meaningful only if the underlying infrastructures are resilient and sustainable.

The question is no longer simply whether health systems can adopt AI. Increasingly, policymakers must ask whether health systems can depend upon it.

A note of caution

Indonesia’s approach warrants cautious optimism.

The intention to establish oversight mechanisms, strengthen ethics review capacities, and pursue sector-specific regulation reflects an appreciation that healthcare requires differentiated governance.

At the same time, important challenges remain.

Broad frameworks must translate into practical guidance. Oversight bodies require expertise and resources. Most importantly, scientific evidence informing policy must remain insulated from both political and commercial pressures.

These are not uniquely Indonesian concerns. They represent some of the defining governance challenges confronting healthcare systems globally.

The real test ahead

Indonesia’s recent announcements are not a blueprint for AI governance. They are, however, an important acknowledgement that governance must evolve alongside technological capability. In a region as diverse as Asia, context matters, institutional realities matter, politics and public expectations matter.

Ultimately, the success of AI will depend less on the sophistication of algorithms and more on the strength of the institutions surrounding them. The measure of responsible AI will not be what these systems can do, but whether societies remain capable of governing them in ways that are equitable, accountable, and anchored in the public interest.

Indonesia deserves recognition not for having all the answers, but for beginning to ask the right questions.