Choices that will define Asia’s AI future are coming into focus

We are fortunate to live at an incredible moment in history. As more and more AI applications have become available, people have turned to them to seek advice, build businesses, express ideas, learn, and improve their health and quality of life. Nowhere is this more clear than in Asia. AI is already turbocharging small businesses here. We commissioned YouGov to survey nearly 4,000 small business owners across nine Asia-Pacific markets, and 84% told us their business already uses AI, with 58% using it at least a few times a week. Of those using it, 76% report a positive impact on productivity and 71% on revenue or sales — against 6% reporting a negative one. More striking is what they say it changes. The vast majority say AI lets them do things that once required significantly more money or resources, and 80% say it helps them compete with much larger companies. It is not only the smallest firms. M1, a Singaporean telecoms operator, rebuilt its phone pre-ordering as a conversational journey on WhatsApp rather than a queue and a call center. Sales rose 55%, and the time customers spent ordering fell by more than 60%. The same is true in science and medicine. JAMSTEC, one of the world’s leading ocean research agencies, used Meta’s open model Llama in its work monitoring how the climate is changing around Japan’s coasts. In South Korea, researchers at Seoul National University use another of our open models to interpret medical scans. Superintelligence is the next chapter in this technology’s evolution, which brings two questions into focus. How do we spread the remarkable benefits of this technology as widely as possible? And how do we make sure it is built safely and behaves in a way that is aligned with what people want? On spreading the benefits, Meta’s aim is for everyone to have access to an exceptionally capable agent that understands you and your goals, whoever you are, wherever you live. That’s why we built our agent Muse, which will be available in Singapore, Japan, South Korea and Australia soon. Think of it like having a super smart sidekick working tirelessly for you, that you have complete control over. And we want everyone to be able to use advanced AI models to create things tailored to the needs of their communities and their societies. But a technology only a few institutions can build is not going to serve everyone. That’s the case for open models, which let developers, universities, countries, and individuals download advanced AI models for free. It’s no coincidence that JAMSTEC and Seoul National University’s medical scanning program were built on Meta’s open models. Open models let people and institutions in Asia fine-tune on local data, run it on domestic infrastructure, and work in their own language, without building a frontier lab from scratch. Build on an open foundation and, if that foundation disappears tomorrow, you still have what you built and can use it without anyone’s permission. Closed models offer access, not sovereignty. If an API disappears, you have nothing. On safety, openness is an advantage rather than a concession. For most categories of risk — cybersecurity above all — the open-source record is clear: broad access lets far more people find and fix problems than any closed process can, and it puts powerful models in the hands of the defenders. By allowing researchers, developers, and policymakers to scrutinize and improve AI models, it helps ensure they have high standards of transparency and accountability. Every lab has both the responsibility and the incentive to train its models carefully. People will not use an agent that is misaligned with them, so trust is fast becoming the capability that separates one model from another. That is why we delayed shipping Muse for several months to concentrate on safety and security. Other labs can do the same. And it’s why we were happy to commit – alongside other leading AI labs – to institute a set of controls to ensure we are building AI responsibly. While some around the world are reacting to fears about AI with regulation that stifles innovation, many governments across Asia are choosing policy grounded in science rather than reaction. Japan’s AI Promotion Act rests on principles, official guidance, and voluntary cooperation rather than rigid enforcement. South Korea has legislated, building its framework around promotion and trust, and concentrating its rules on the highest-impact systems. Singapore launched the world’s first Model AI Governance Framework for Agentic AI and is leading the development of Guides for AI Governance and Ethics at ASEAN. We believe this future should be for everyone, with superintelligence available broadly and people building for their communities and their societies with open models. That won’t happen by accident. The choices being made now about how to regulate this remarkable technology will be hard to unmake. Policymakers should meet this extraordinary moment with optimism. Demand openness, demand responsibility, and demand access for everyone. Joel Kaplan is Meta’s chief global affairs officer. He will be speaking at the 2026 Milken Institute Asia Summit.