Asia loves AI. Governing it is a different story.

Southeast Asia leads the world in AI optimism and government trust–but Stanford's 2026 data shows the region's responsible AI infrastructure is still catching up to its own enthusiasm.

There is a gap running through Stanford University's 2026 AI Index Report that no amount of optimism data can paper over. The region most enthusiastic about artificial intelligence–most trusting of its governments, most convinced the technology will change lives–is also the region where the infrastructure to govern that technology responsibly remains a work in progress.

That is the tension at the centre of the report's public opinion findings, published last week by Stanford's Institute for Human-Centered Artificial Intelligence. And for Asia, it is the finding worth sitting with.

The numbers are striking

In Malaysia, Thailand, Indonesia, and Singapore, over 80% of respondents say AI will profoundly change their lives in the next three to five years, according to Ipsos' 2025 AI Monitor survey, cited in the report. That figure alone sets the region apart. Globally, nervousness about AI has been rising with 52% of respondents worldwide saying AI products make them nervous, up two percentage points from 2024.

In Southeast Asia, that nervousness sits well below the global average. China and Indonesia show the highest levels of excitement of any countries surveyed, with nervousness below 50%.

Malaysia recorded the largest single-year increase in AI optimism of any country in the 30-nation survey, up nine percentage points from 2024.

Trust in government regulation follows the same pattern. Singapore leads all 30 surveyed countries at 81%, meaning 81% of Singaporeans trust their government to regulate AI responsibly. Indonesia sits at 76%, Malaysia at 73%, Thailand at 70%. The global average is 54%. The United States, for context, sits last at 31%.

These are not marginal differences. Southeast Asia's trust advantage over Western markets is substantial, and it matters because public trust in government is what creates the political conditions for meaningful AI regulation to be enacted and accepted.

Where the governance picture gets complicated

The report's responsible AI chapter, which draws on a survey conducted jointly with McKinsey across multiple regions and industries, introduces a four-point maturity scale for how organizations are implementing responsible AI. Level one means foundational practices exist. Level four means comprehensive and proactive practices are fully operational.

In 2025, the global average was 2.3. Asia-Pacific registered 2.5, the highest regional score, and an improvement from 2.2 in 2024. At first reading that sounds encouraging. But a score of 2.5 places the region squarely in the "integrating" band, meaning practices are being embedded but are not yet operational at the level the framework defines as complete.

The barriers are consistent with what the region's enterprise AI conversation already suggests. Knowledge and training gaps were cited as the top obstacle to responsible AI implementation globally, by 59% of respondents, up from 51% in 2024. Technical limitations rose to 38%, from 32%. For a region where public enthusiasm for AI is outpacing institutional readiness, the skills gap is the most immediate structural constraint.

There is also a more fundamental challenge the report surfaces. Responsible AI is not a single dimension. It covers fairness, transparency, safety, privacy, accountability, and more–and the report finds that improving one dimension can actively degrade another. Gains in safety can reduce accuracy.

Gains in privacy can reduce fairness. There is no established global framework for navigating these trade-offs, and for several dimensions, the standardized data needed to track progress over time does not yet exist.

What the region is building toward

National AI strategies are expanding across Asia. The report notes that state-backed investments in AI supercomputing are rising and that developing economies are increasingly asserting ambitions for domestic control over AI ecosystems. Yet model production remains concentrated in the United States and China.

Most of Southeast Asia is deploying AI systems built elsewhere, which means the governance frameworks being developed need to account for dependency on foreign AI infrastructure, not only domestic deployment decisions. That dependency question is not unique to Asia, but it is more acute here.

The region's data centre buildout is accelerating with Malaysia alone accounting for a significant share of planned regional capacity, even as the models running on that infrastructure, and the companies setting the terms of their use, remain largely outside the region's regulatory reach.

Globally, the EU is trusted more than the United States or China to regulate AI effectively. Across 25 countries surveyed by Pew Research Centre in 2025, a median of 53% trusted the EU on AI regulation, compared to 37% for the United States and 27% for China. For Asian governments building their own frameworks, the EU's approach is increasingly the reference point, even as the timeline for getting there remains compressed.

The optimism is real. The work ahead of it is too.

Stanford's report does not position Southeast Asia as unprepared. The data reflects a region with genuine institutional advantages, high public trust, strong enthusiasm, and improving organizational maturity, building its governance capacity under real pressure of time.

What the numbers make clear is that enthusiasm and infrastructure are not the same thing. Asia's relationship with AI is, by the data, the most confident in the world. Whether the governance frameworks can keep pace with that confidence is the question the 2026 AI Index leaves open.