TOKYO: Since ChatGPT stunned the world three years ago with the powers of generative AI, countries have grappled with how to govern the rapidly developing technology.

As Vietnam’s artificial intelligence law goes into effect on Sunday, this news agency takes a look at regulation efforts around the globe:

EU: first mover

The European Union is considered a trailblazer, having adopted in 2024 what it calls “the world’s first comprehensive AI law” penalised with heavy fines. The law takes a risk-based approach: if a system is high-risk, a company will have a stricter set of obligations to fulfil before being authorised in the EU.

These landmark rules have faced pushback from Washington under President Donald Trump, but also from businesses and governments at home that complain they could hamper growth.

The EU bowed to pressure last year and proposed changes including partially delaying the law’s application in a move it says will help European companies compete globally.

The law will now be fully applicable in 2027, but the EU already allows regulators to ban systems deemed to pose unacceptable risks. That could include “social scoring” systems that lead to discrimination by classifying individuals or groups based on social behaviour or personal traits.

US: land of the free?

The United States, home to ChatGPT maker OpenAI, chip titan Nvidia and tech giants like Google, is not keen on enacting new rules.

Vice President JD Vance has warned against “excessive regulation” that “could kill a transformative sector”.

And at a major AI summit in New Delhi in February, the US delegation head said the country “totally” rejects global governance of AI.

Some states have taken matters into their own hands, although the White House is trying to find ways to stop this from happening.

California enacted a first-of-its-kind law in October requiring AI chatbot operators to implement safeguards such as referring people who express thoughts on suicide to crisis services.

Asia: new laws

A wide-ranging law took full effect in South Korea in January, requiring companies to tell users when products use generative AI. It also says they must clearly label content, including deepfakes, that cannot readily be differentiated from reality.

Places like Taiwan and Japan are taking a lighter touch on AI regulation, shying away from penalties in favour of voluntary guidelines promoting innovation.

China — racing to challenge US dominance in the technology — has its own complex and evolving set of guardrails. “What stands out about China is how regulated it is, despite the innovation happening,” Seth Hays, author of the Asia AI Policy Monitor newsletter, said.

“For example, firms need to register their models with the government, and there are stricter content moderation, labelling and user verification requirements not seen elsewhere.”

Global debate

Many other countries, from Brazil to the United Arab Emirates, are implementing AI frameworks that can roughly be divided into risk-based rules like the EU’s, or pro-innovation guidelines. At the New Delhi summit, 91 countries and organisations called for “secure, trustworthy and robust” AI. But their declaration, signed by the United States and China, was criticised by AI safety campaigners for being too generic to protect the public.

A 40-member United Nations expert panel has also been established to work towards “science-led governance” of the technology, the global body’s chief Antonio Guterres has said.

Published in Dawn, March 1st, 2026