Policymakers set to prepare more AI rules
text size

Policymakers set to prepare more AI rules

Countries likely to look to EU's AI Act

AI standards are an important part of building consumer acceptance, according to researchers.
AI standards are an important part of building consumer acceptance, according to researchers.

Policymakers will draw up more new artificial intelligence (AI) rules to prevent the violation of AI ethics and promote AI development through good governance and planned guidelines for generative AI.

Meanwhile, legal experts expect that many countries, including Thailand, will develop comprehensive AI legal frameworks following the European Union's AI Act, which is considered to be the world's first comprehensive AI law.

The European Parliament approved the law last month, with the aim of constraining the risks of AI.

Putchapong Nodthaisong, secretary-general of the National Digital Economy and Society Commission, told the Bangkok Post the commission has AI ethic guidelines in place.

By the end of May, its digital economy promotion subcommittee will convene to consider drafting AI regulations to enforce upon those who violate AI ethics as part of its supervision of AI developers and AI service providers.

Chaichana Mitrpant, executive director of the Electronic Transactions Development Agency (ETDA), told the Bangkok Post that the agency and Thammasat University had jointly studied AI regulations in Europe and Brazil for almost two years and found the continent and country, respectively, have managed to prohibit high risk AI.

They also promoted AI within a sandbox and offered incentives to accelerate the development of AI.

He added that Thai regulators feel there is no need to urgently impose strong comprehensive AI regulations, while startups prefer self-regulation in the future and consumers are concerned about scams and the use of deepfake technology.

Mr Chaichana said Thailand could learn from other major countries regarding the impact of the enforcement of AI regulations.

The ETDA currently has an AI governance clinic to assess the level of risk, make recommendations and monitor the use of AI-related projects.

It is now drafting generative AI governance guidelines for AI developers, so they can avoid AI hallucinations and avoid intellectual property infringements.

"AI hallucinations" refers to the AI generation of incorrect or misleading output.

The ETDA has also drafted a law on the application of AI with good governance. The key components of this draft law include an AI innovation testing centre, an important mechanism for AI research and development that requires the testing of algorithmic systems.

The other core component is information sharing. It requires the ETDA to promote, support and provide assistance to government agencies, the private sector and the general public in promoting information sharing to develop AI innovation.

The draft law also requires the creation of AI standards by setting and certifying standards and creating an AI standard logo. All these would provide necessary information to consumers in order to make informed decisions when choosing products or services.

The law also requires the ETDA to prepare criteria and methods for assessing risks from the use of AI in order to prepare various measures or methods for dealing with dangers that may occur in the future from high risk AI.

The law will also govern the standardisation of contracts between service providers and users of AI products or services. The contracts must provide greater clarity in terms of legal relations between service providers and users. This is to prevent problems that may arise from not knowing or not understanding the complex systems of AI.

Apivadee Piyatumrong, a researcher at the National Electronics and Computer Technology Center, said AI standards are an important part of building consumer acceptance.

Pattaraphan Paiboon, an IP tech partner of Baker & McKenzie Ltd, said it is likely that many countries will follow in the footsteps of the EU's AI Act and come up with their own AI act.

For Thailand, the current laws may not be sufficient to support or regulate AI due to the uniqueness of the technology. However, as Thailand is at the early stage of implementing AI for businesses and services, it might be too early to impose comprehensive AI regulations.

Ms Pattaraphan added that the government also has an overarching national AI development roadmap, requiring that an AI law has to be enacted by 2027.

The government will likely study the effects of the EU's AI Act and will adopt and adjust draft AI laws for Thailand accordingly.

Dhiraphol Suwanprateep, an IP tech of counsel at Baker & McKenzie, said the government should mainly aim to support and promote the use and development of AI and AI businesses, similar to China's regulations on the promotion of the AI industry in the Shenzhen Special Economic Zone, rather than strictly regulating and hindering AI development in Thailand.

If any AI regulation is necessary, Thailand may focus on regulating generative AI first, similar to China's interim measures for the management of generative AI services, instead of issuing comprehensive AI legislation, he said.

Given that the draft AI laws are still in the early development stage, stakeholders may voice their opinions to the regulators to help shape AI laws for Thailand in the future, Ms Pattaraphan said.

Christina Montgomery, vice-president and chief privacy and trust officer at IBM, said the EU's AI Act is essentially a product safety law, which means that if companies develop or deploy AI products in the EU, businesses fall under the scope of its risk-based rules.

"That is why IBM encourages companies everywhere to understand and prepare for these incoming regulations," said Ms Montgomery.

Do you like the content of this article?